SUMMARY TECHNICAL REPORT OF THE NATIONAL DEFENSE RESEARCH COMMITTEE This document con taimpjm formation affecting the national defense of't>ie United States within the meaning of the EspiomAcAct, 50 U. S. C., 31 atid 32, amended. Its tran»ission or the revelatiop contents in any manner to amyiinauthorizedPperson is prohibited // This volume is classified RESTRICTED1 in accordance with, purity regula- tions of thj»Var and Navy Departments because certain chapters contain ma- terial wh«i was RESTRICTED at; the date of printing^Other chapters may have lyi a lower Classification afr none. The to consult the Navy agencies listed 'on the reverse of this* page for the current' clarification of any matcmT Manuscript and illustrations for this volume were prepared for publication by the Summary Reports Group of the Columbia University Division of War Research under contract OEMsr-1131 with the Office of Scientific Research and Development. This vol- ume was printed and bound by the Columbia University Press. Distribution of the Summary Technical Report of NDRC has been made by the War and Navy Departments. Inquiries con- cerning the availability and distribution of the Summary Tech- nical Report volumes and microfilmed and other reference ma- terial should be addressed to the War Department Library, Room 1A-522, The Pentagon, Washington 25, D. C., or to the Office of Naval Research, Navy Department, Attention: Reports and Doc- uments Section, Washington 25, D. C. Copy No. 246 This volume, like the seventy others of the Summary Technical Report of NDRC, has been written, edited, and printed under great pressure. Inevitably there are errors which have slipped past Division readers and proofreaders. There may be errors of fact not known at time of printing. The author has not been able to follow through his writing to the final page proof. Please report errors to: JOINT RESEARCH AND DEVELOPMENT BOARD PROGRAMS DIVISION (STR ERRATA) WASHINGTON 25, D. C. A master errata sheet will be compiled from these reports and sent to recipients of the volume. Your help will make this book more useful to other readers and will be of great value in preparing any revisions. SUMMARY TECHNICAL REPORT OF THE APPLIED PSYCHOLOGY PANEL, NDRC VOLUME 2 HUMAN FACTORS IN MILITARY EFFICIENCY TRAINING AND EQUIPMENT OFFICE OF SCIENTIFIC RESEARCH AND DEVELOPMENT VANNEVAR HUSH, DIRECTOR NATIONAL DEFENSE RESEARCH COMMITTEE JAMES B. CONANT, CHAIRMAN APPLIED PSYCHOLOGY PANEL CHARLES W. BRAY, CHIEF WASHINGTON, D. C., 1946 NATIONAL DEFENSE RESEARCH COMMITTEE James B. Conant, Chairman Richard C. Tolman, Vice Chairman Roger Adams Frank B. Jewett Karl T. Compton Army Representative1 Navy Representative2 Commissioner of Patents3 Irvin Stewart, Executive Secretary 1 Army representatives in order of service: 2 Navy representatives in order of service: Maj. Gen. G. V. Strong Maj. Gen. R. C. Moore Maj. Gen. C. C. Williams Brig. Gen. W. A. Wood, Jr. Col. L. A. Denson Col. P. R. Faymonville Brig. Gen. E. A. Regnier Col. M. M. Irvine Rear Adm. H. G. Bowen Capt. Lybrand P. Smith Rear Adm. J. A. Purer Rear Adm. A. H. Van Keuren 3 Commissioners of Patents in order of service: Commodore H. A. Schade Col. E. A. Routheau Conway P. Coe Casper W. Ooms NOTES ON THE ORGANIZATION OF NDRC The duties of the National Defense Research Committee were (1) to recommend to the Director of OSRD suitable projects and research programs on the instrumentalities of warfare, together with contract facilities for carry- ing out these projects and programs, and (2) to ad- minister the technical and scientific work of the con- tracts. More specifically, NDRC functioned by initiating research projects on requests from the Army or the Navy, or on requests from an allied government trans- mitted through the Liaison Office of OSRD, or on its own considered initiative as a result of the experience of its members. Proposals prepared by the Division, Panel, or Committee for research contracts for performance of the work involved in such projects were first reviewed by NDRC, and if approved, recommended to the Director of OSRD. Upon approval of a proposal by the Director, a contract permitting maximum flexibility of scientific effort was arranged. The business aspects of the con- tract, including such matters as materials, clearances, vouchers, patents, priorities, legal matters, and admin- istration of patent matters were handled by the Execu- tive Secretary of OSRD. Originally NDRC administered its work through five divisions, each headed by one of the NDRC members. These were: Division A—Armor and Ordnance Division B—Bombs, Fuels, Gases, & Chemical Prob- lems Division C—Communication and Transportation Division D—Detection, Controls, and Instruments Division E—Patents and Inventions In a reorganization in the fall of 1942, twenty-three administrative divisions, panels, or committees were created, each with a chief selected on the basis of his outstanding work in the particular field. The NDRC members then became a reviewing and advisory group to the Director of OSRD. The final organization was as follows: Division 1—Ballistic Research Division 2—Effects of Impact and Explosion Division 3—Rocket Ordnance Division 4—Ordnance Accessories Division 5—New Missiles Division 6—Sub-Surface Warfare Division 7—Fire Control Division 8—Explosives Division 9—Chemistry Division 10—Absorbents and Aerosols Division 11—Chemical Engineering Division 12—Transportation Division 13—Electrical Communication Division 14—Radar Division 15—Radio Coordination Division 16—Optics and Camouflage Division 17—Physics Division 18—War Metallurgy Division 19—Miscellaneous Applied Mathematics Panel Applied Psychology Panel Committee on Propagation Tropical Deterioration Administrative Committee NDRC FOREWORD As EVENTS of the years preceding 1940 re- vealed more and more clearly the serious- ness of the world situation, many scientists in this country came to realize the need of organ- izing scientific research for service in a national emergency. Recommendations which they made to the White House were given careful and sym- pathetic attention, and as a result the National Defense Research Committee [NDRC] was formed by Executive Order of the President in the summer of 1940. The members of NDRC, appointed by the President, were instructed to supplement the work of the Army and the Navy in the development of the instrumentalities of war. A year later, upon the establishment of the Office of Scientific Research and Development [OSRD], NDRC became one of its units. The Summary Technical Report of NDRC is a conscientious effort on the part of NDRC to summarize and evaluate its work and to present it in a useful and permanent form. It comprises some seventy volumes broken into groups cor- responding to the NDRC Divisions, Panels, and Committees. The Summary Technical Report of each Divi- sion, Panel, or Committee is an integral survey of the work of that group. The first volume of each group’s report contains a summary of the report, stating the problems presented and the philosophy of attacking them, and summarizing the results of the research, development, and training activities undertaken. Some volumes may be “state of the art” treatises covering subjects to which various research groups have contributed information. Others may contain descriptions of devices developed in the labora- tories. A master index of all these divisional, panel, and committee reports which together constitute the Summary Technical Report of NDRC is contained in a separate volume, which also includes the index of a microfilm record of pertinent technical laboratory reports and ref- erence material. Some of the NDRC-sponsored researches which had been declassified by the end of 1945 were of sufficient popular interest that it was found desirable to report them in the form of monographs, such as the series on radar by Division 14 and the monograph on sampling inspection by the Applied Mathematics Panel. Since the material treated in them is not du- plicated in the Summary Technical Report of NDRC, the monographs are an important part of the story of these aspects of NDRC research. In contrast to the information on radar, which is of widespread interest and much of which is released to the public, the research on subsurface warfare is largely classified and is of general interest to a more restricted group. As a consequence, the report of Division 6 is found almost entirely in its Summary Technical Report, which runs to over twenty volumes. The extent of the work of a Division cannot therefore be judged solely by the number of volumes devoted to it in the Summary Technical Report of NDRC: account must be taken of the monographs and available reports published elsewhere. The Applied Psychology Panel, under the di- rection first of W. S. Hunter and later of C. W. Bray, comprised a small group of psychologists and personnel specialists whose responsibility was to aid in refining and standardizing Army and Navy personnel procedures. The Panel de- vised selection and classification tests; it devel- oped training methods; it improved the design of much equipment. The work of the Panel proved that it is as important to get the right man for a military job as it is to get the right ammunition for his gun. The achievements of the Applied Psychology Panel cannot be measured in quantitative terms. But one can, for example, estimate with cer- tainty that the tests devised to eliminate the emotionally unfit from induction prevented the wrecking of many lives and the fruitless ex- penditure of much time, effort, and money; and one can know surely that many lives were saved as the result of the one study alone which showed that the best night lookouts at sea were four times as proficient as the poorest. The Summary Technical Report of the Panel, prepared under the direction of the Panel Chief and authorized by him for publication, is a rec- ord of scientific accomplishment and of zealous effort by able men working to increase the ef- fectiveness of the nation’s military manpower in time of national peril. The members of the Panel have our gratitude. Vannevar Bush, Director Office of Scientific Research and Development J. B, Conant, Chairman National Defense Research Committee FOREWORD The applied psychology panel of NDRC was organized in June 1942 in order to mobilize civilian assistance for a program of application of psychology to military problems. Outstanding at the time was the problem of assigning men to duty in terms of aptitude. Aptitude testing was the primary contribution of psychology to military efficiency in World War I. In World War II, psychologists were engaged at an early date to apply this specialty to the problems of mass classification. In the peacetime years between the two wars, many research studies of human learning were conducted. The principles of learning developed in the laboratory, and their application to mass training of industrial personnel, were obviously relevant to military training. Hence, when the Applied Psychology Panel was formed, it was asked to assist in military training as well as in aptitude testing. With the extraordinary freedom permitted to scientists by the NDRC organization, the psychologists of the Applied Psychology Panel branched into a third field: the design and op- eration of military equipment. Materiel must be operated by men. In wartime it must be operated by half-trained men, by men who are not confident of themselves, by men who make mistakes. Hence the design and operation of equipment is not a problem of materiel alone. It is a problem of man and machine in combina- tion. To this problem the psychologists of the Applied Psychology Panel made contributions. As the war progressed, the Panel’s work on aptitude, on training, and on equipment became more unified. If aptitude tests were improved, the men sent to a given school could be more readily trained; operating procedures and equipment could be more complex. As training became more specific, general verbal ability tests became less useful and special aptitude tests more necessary. With high quality train- ing, complex materiel was more often used and less often misused. When materiel was designed in terms of the needs of the average soldier or sailor, operation became simpler and the re- quirements for selection and training were re- duced. Not only were the problems of aptitude, training, and equipment seen to be interrelated, but the same research method was found to be useful in each of the three problems. In all three cases, objective measures of human perform- ance were necessary as criteria to use in evalu- ating alternative procedures. Such criteria were needed to measure the knowledge and skill of men who were selected by alternative aptitude tests, of men who were trained on different synthetic devices, of men who were following opposed doctrines of use of equipment, or of men who were serving as human guinea pigs in the comparison of alternative designs of equipment. Thus the projects of the Applied Psychology Panel tended to become centers for a coordinated approach to the problems of using the devices of war. The results of these coordinated studies of the problems of military psychology are de- scribed in the APP Summary Technical Report under the general title, Human Factors in Mili- tary Efficiency. Volume 1 is subtitled Aptitude and Classification. This volume is subtitled Training and Equipment. The first half of this volume describes the research on the training of various specialists. The second half describes research on the design and operation of the equipment for which training was necessary. The research reported in the Summary Tech- nical Report resulted in an increase in the pro- ficiency of soldiers and sailors in combat. The application of psychological techniques in- creased the speed, for example, with which messages were sent over radio and interphone in the noisy conditions of combat. The accuracy of antiaircraft gunners was raised. The ability of radar operators to read the position and range of enemy targets became greater. Am- phibious personnel were better able to recognize beach markers. Yet the positive values of psychological re- search were limited in World War II by condi- tions which can now be corrected. Greater value could have been obtained if research studies had been activated earlier. In many cases psy- chological research began too late to be of maximal effectiveness. Several examples may VIII FOREWORD be mentioned from the experience of the Ap- plied Psychology Panel. In the early years of the war in Europe, at Pearl Harbor, and in the further reaches of the Pacific, the need for antiaircraft defense was made obvious. But not until 1942 and 1943 was the Panel asked to help improve the efficiency of antiaircraft gunners. In 1942 and 1943 the pri- ority of ground warfare increased. But it was 1944 before a project on human errors in the use of field artillery was activated. It was 1945, after the B-29 airplane had seen combat, before psychological research was requested in order to help formulate military requirements for aerial gunnery equipment. In each instance months were required to obtain useful re- sults. In each instance significant research results were obtained. In each instance these results would have had greater military value if the need had been anticipated and the research requested before instead of after that need be- came acute. In the future, new devices must be considered in terms of personnel requirements before pro- duction begins. New devices always impose a strain on personnel. Officers in training camps or in the field cannot be expected to produce efficient fighting teams when the officers them- selves are unfamiliar with the new devices which must be used. The psychological prob- lems can be anticipated, and many military problems eliminated before they occur, if quali- fied psychologists are encouraged to study new devices which are at the blueprint, the mock-up, or the reproduction stage of development. If psychological research on selection, train- ing, and equipment is concentrated on prepro- duction models, the results will appear in time to be of maximal usefulness. Standard operat- ing procedures will be worked out by the time the new equipment is ready for distribution. Training plans can be developed before actual training begins. Selection requirements will be known before men are selected for training on the new equipment. Postponing psychological studies until after general use has demonstrated the shortcomings or difficulties of a piece of equipment can ac- complish only corrective measures. Anticipating those shortcomings by subjecting equipment to psychological study during its early develop- ment will lead to better selection, better train- ing, better operating procedures, and better equipment. Charles W. Bray Chief, Applied Psychology Panel PREFACE The app summary technical report is a systematic account of the work done under the direction of the Applied Psychology Panel of the National Defense Research Committee. It includes work done before the Panel’s forma- tion when the same projects and personnel were under the supervision of the National Research Council’s Committee on Service Personnel— Selection and Training. Volume 1 describes se- lection and classification of military personnel; Volume 2 describes military training and the human factors involved in the design and op- eration of military equipment. In each of these three fields—selection and classification, train- ing, and the design and operation of military equipment—the work actually done and the effects of that work on military practice are described. Chapter 1 summarizes this volume. The re- maining chapters can conveniently be divided into four groups: Chapters 2 to 12 relate the specific efforts of Applied Psychology Panel projects to improve the training given to a number of different types of military special- ists. Chapters 13 to 17 discuss the psychological principles and the methods employed in organ- izing mass instruction, developing instructional aids, evaluating synthetic trainers, and meas- uring the achievement of men in training. Chapters 18 to 23 describe the work done by Panel projects on improving the design and operational procedures for a number of types of military equipment. Chapters 24 and 25 pre- sent psychological principles for designing mil- itary equipment and for developing the most efficient operating procedures for that equip- ment. As was true of Volume 1, cross references are given by means of section numbers, for example 11.3.5, in which the 11 refers to the chapter, 3 to the third major division of Chap- ter 11, and 5 to the fifth section of the third division of Chapter 11. Commonly used abbre- viations are explained in a glossary at the end of the volume. In writing this final summary a few tables were recalculated from the original reports. In no case were the changes large enough to alter conclusions or recommendations. The Applied Psychology Panel has had help from many sources in preparing this final ac- count of its work. To each of these we express our thanks. The Army, the Navy, and the Applied Psy- chology Panel Contractors have provided pho- tographs to illustrate many of the devices and procedures discussed. The author of each chapter is named in the table of contents and at the beginning of the chapter. Though all chapters are based upon the original reports prepared by the contrac- tors,'1 many of them were finally written by the editor. Approximately half of the chapters were written by men who spent the war years in the field, working on problems their chapters sum- marize. The Panel and the editor are very ap- preciative of the time and effort devoted by these men to writing first-hand accounts of their work. We are particularly indebted to William E. Kappauf, Jr., who wrote seven of the chapters in this volume. As in Volume 1, credit is not given to the individual psychologists who actually made the contributions here reported. Their specific con- tributions can be discovered only by studying the original reports listed in the bibliographies. The Applied Psychology Panel expresses its sincere appreciation to these men for their in- dividual contributions and for their effective teamwork. These two volumes record their achievement. Dael Wolfle Editor a A complete bibliography of reports prepared by Applied Psychology Panel contractors is contained in Final Report and Bibliography of the Applied Psy- chology Panel, Charles W. Bray, (DSHD 6668, June 30, 1946. This report gives the title and other identifying data of each report. CONTENTS CHAPTER 1 Summary by Dael Wolfie 1 PAGE PARTI TRAINING OF SPECIALIZED MILITARY PERSONNEL 2 Search-Radar and Bombing-Radar Operators by Donald B. Lindsley 8 3 Training Trackers for Antiaircraft Fire Con- trol by William C. Biel and William E. Kap- pauf/jr. 19 4 Training Personnel in Range Determination by William E. Kappauf, Jr 41 5 d raining the B-29 Gunner by Charles W. Bray 57 6 Night Lookouts by Dael Wolfie .... 63 7 Training Navy Gunners and Engine Room Crews by Bernard J. Conner 79 8 Training Winch Operators by Dael Wolfie . 83 9 Training Amphibious Craft Crews by Dael Wolfie 86 10 Training Navy Telephone Talkers by Dael Wolfie 90 11 Voice Communication Training in the Army Air Forces by Dael Wolfie 98 12 Training Radio Operators by Dael Wolfie . 106 PART II PSYCHOLOGICAL PRINCIPLES IN MILITARY TRAINING 13 Psychological Principles in Military Training by Dad Wolfie 124 14 Job Analysis of Duties to Be Learned by BernardI J. Cornier 127 15 Course Outlines and Lesson Plans by Bernard J. Conner 132 16 The Use and Design of Synthetic d rainers for Military Training by Dael Wolfie .... 140 17 Measuring the Effects of Training by Norman Frederiksen 155 CONTENTS CHAPTER PAGE PART III THE DESIGN AND OPERATION OF SPECIAL TYPES OF MILITARY EQUIPMENT 18 Antiaircraft Directors and Guns by William E. Kappauf, Jr 179 19 Field Artillery Equipment by William E. Kappauf, Jr 206 20 The Design and Operation of B-29 Gunsights by Charles W. Bray 219 21 The Operation of Radar Equipment by Donald B. Lindsley 234 22 Stereoscopic Rangefinders and Heightfinders by William E. Kappauf, Jr 251 23 Standard Procedures in Voice Communica- tion by Dael Wolfle 270 PART IV GUIDING PRINCIPLES FOR FUTURE EQUIPMENT DEVELOPMENT 24 Principles of Good Equipment Design by William E. Kappauf, Jr 284 25 The Development of Standard Operating Pro- cedures by William E. Kappauf, Jr. . . . 296 Glossary 321 Bibliography 323 OSRD Appointees 349 Contract Numbers 351 Service Project Numbers 353 Index 355 Chapter 1 SUMMARY By Dael Wolfie 11 INTRODUCTION This CHAPTER is a summary of the work of the Applied Psychology Panel, NDRC, on problems of military training and equipment, the work reported in the remaining chapters of this volume. This summary is divided into four sections which parallel the four major divisions of the volume. Section 1.2 summarizes specific studies of military training. Section 1.3 out- lines the psychological principles of learning and general methods of improving training. Section 1.4 describes specific studies directed toward improving the design of military equip- ment and the development of better operating procedures for such equipment. Section 1.5 briefly reviews the psychological principles in- volved in equipment design and operation. The subtitles of the following sections in- clude the numbers of the chapters summarized in each. A rapid reading will give the location and plan of organization of the main research findings of the entire volume. 12 TRAINING OF SPECIALIZED MILITARY PERSONNEL Chapters 2 to 13 deal with the application of general psychological principles to specific learning situations and pieces of equipment. In each case, work was directed toward the achievement of one or more of three objectives: to make the training more specifically directed toward the actual duties to be performed; to include better measures of progress and achievement; and to make fuller use of estab- lished principles of learning. 1-2-1 Training Search-Radar and Bombing- Radar Operators (Chapter 2) Surveys of radar operator training schools led to a number of specific suggestions for im- proving the instruction given. The task of the search-radar operator is in general the same, even though there are wide differences in equipment and in special pro- cedures. Common elements in the task were studied: the tactical use of radar equipment; specific procedures, such as calibration and tuning; manipulation of controls; interpreta- tion of scope phenomena; and the reading of ranges and bearings. These studies resulted in the development of synthetic trainers, achieve- ment measures, and instructional aids, includ- ing the following examples: several mechanical- optical plan position indicator [PPI] trainers to aid in the training of search-radar operators; a flash-reading method to train men in the rapid identification and interpretation of scope phenomena; an H2X film trainer and radar scope movies for ground training and for use in briefing radar bombing operators for train- ing missions. Reductions in average bombing errors of from 1,000 to 1,500 feet were secured by giving trained bombardiers an additional 150 hours of practice in operation of the main scope. 122 Training Trackers for Antiaircraft Fire Control (Chapter 3) Checksights were developed for use on a number of different guns. It was demonstrated that trackers learn more rapidly when trained with checksights than when trained without them. Checksights were found to be the most generally useful objective measure of the track- er’s skill. A phototube scoring device to replace the checksight observer was developed too late in the war to be used. Its reliability and effec- tiveness were, however, demonstrated. Learning records showed that once men were adequately trained in tracking they needed re- fresher drills no more often than once a month. Three on-target tracking trainers were de- veloped: a trainer which simulated the task of optical tracking with the stereoscopic heightfinder; a trainer which simulated the two-man tracking task in the gun director Mark 37; and a trainer which simulated the two-man tracking task on the director M7. The 1 2 SUMMARY last one became known as the Tufts tracking trainer. It had validity as both a selection device and a training device. Several manuals giving instructions for the use of gunnery trainers and the operation of fire-control equipment were written. Training Personnel in Range Determination (Chapter 4) Projects of the Applied Psychology Panel prepared training manuals for instructing Army personnel in heightfinder operation. The same material was in large part adapted for the training of Navy personnel in rangefinder operation. A number of training aids were de- veloped for both Services. The rate of improvement during training in heightfinder and rangefinder operation was studied in a variety of situations. About 2,700 trials were found necessary on the M2 trainer and about 3,200 on the Eastman trainer before men reached a plateau. Men who had had this amount of practice on a trainer reached a plateau in their ability to range on aerial tar- gets in about 500 trials. Unaided estimation of the opening range for 20 mm fire, 1,700 yards, was found to be superior to stadiametric ranging with the gun- sight Mark 14. Training on the firing line with the reticle of the gunsight Mark 14 was found to be superior to training on the mirror range estimation trainer device 5C-4 in learning sta- diametric ranging. Practice in aerial target range estimation increased the percentage of estimates within 15 per cent of true range from about 25 to about 45 per cent. Relatively frequent refresher drills were found necessary, however, to keep men at this level of performance. The accuracy of estimating the range of aerial targets is approximately the same as the accuracy of estimating the range of ground targets. 1.2.4, Training the B-29 Gunner (Chapter 5) Applied Psychology Panel project personnel studied the scoring of gun camera film, a check- sight for B-29 gunners, and the effects of coaching in a training program. A new scoring method for gun camera film was developed which required less than one-third as many man hours as the standard Army method. The checksight provided unreliable scoring but probably motivated the men to better perform- ance. The data of the coaching experiment in- dicated that gunners tended to over-range at the beginning and to under-range at the end of an attack. i-2-5 Night Lookouts (Chapter 6) A shipboard study of the performance of night lookouts on duty in an Atlantic convoy provided specific information on the average ability and the variation in ability of night lookouts to spot targets. The best lookouts were able to see a ship nearly four times as far away as the poorest ones. These differences were not due to differences in night vision as measured by the adaptometer. Two night vision trainers and a training ex- ercise in the use of binoculars at night were constructed and submitted to the Navy. Assist- ance was given in the preparation of literature for the training of lookouts. 126 Training Navy Operating Crews (Chapters 7 and 8) Lesson plans, course outlines, and instruc- tional manuals were written for the .50 caliber machine gun, for the 20 mm, 40 mm, 8"/50, 4"/50, anci 5"/38 guns, and for the main bat- tery of large ships. Similar teaching aids were written for the instruction of operating engi- neers in training for duty aboard a number of types of Navy ships. Paralleling the work on lesson plans for gun- nery instruction was a series of studies on gunnery trainers and lesson plan materials for their use. Courses entitled How to Teach Gunnery and How to Teach Engineering were developed and given to a number of classes of instructors. A trainer consisting of a miniature electric winch was constructed to help train hatchmen and winchmen for duty on APA’s and AKA’s. TRAINING OF SPECIALIZED MILITARY PERSONNEL 3 1-2,7 Training Amphibious Craft Crews (Chapter 9) Project personnel assisted the Amphibious Training Command in the development and improvement of training courses for the crews of amphibious craft. Job analyses were made of amphibious en- listed billets and a questionnaire was admin- istered to personnel returning from combat zones in order to secure information about the effectiveness of various methods of training, classification, or ship performance in actual combat. The training studies and the curricula de- veloped were organized into a systematic over- all program described in the Amphibious Train- ing Command Training Manual. Voice Communication Training (Chapters 10 and 11) The high noise levels encountered in military airplanes and aboard ship frequently made normal speech unintelligible over telephone and other communication circuits. Special training courses were developed in order to increase intelligibility, A Telephone Talkers' Manual, a manual for instructors, and phonograph records were de- veloped for Navy telephone talkers. Experi- mental investigations of the best methods of instruction and of the value of various parts of the course and a survey of the training given to telephone talkers in a number of installa- tions led to improved standardized instruction. The training ashore was integrated with sub- sequent training on board ship. Three courses were developed for the sub- marine service: a basic course to teach general skills; an intermediate course to give each man proficiency in the use of the phraseology re- quired by his assignment aboard ship; and an advanced course to drill the entire crew as a combat team in the coordinated use of the com- munication circuits. In the Army Air Force, training methods were devised and instructors’ handbooks and students’ manuals written for courses designed to teach personnel to speak so that their mes- sages would be more intelligible. Special courses were written for the training of each type of aircrew specialist and for the training of formed crews of heavy and very heavy bombers. Training Radio Operators (Chapter 12) The code-voice method of teaching basic code was developed on the basis of established principles of learning. It was shown to be superior to former methods and was adopted by the Army. Four hours of daily drill were shown to pro- duce as rapid learning as 7. The distribution of these 4 hours within a day was found to be unimportant. Increasing the variety of drill materials produced more rapid learning and decreased boredom in both students and in- structors. One hour a day of practice in copying hand-sent dear-text practice material produced a small improvement in ability to copy cipher accurately. Giving men practice in copying code through various types of interfering noises did not diminish their ability to copy clear code and led to a moderate improvement in ability to copy code through interference. A trainer to aid men in learning to send cor- rectly was developed. It consisted of a type- writer which was controlled by electronic cir- cuits in such a way that Morse code characters were transcribed by the typewriter as ordinary letters. Correct sending appeared immediately on the typewriter as correct copy. Errors in sending appeared as errors and informed the student immediately of the nature of the mis- take he had made. A monograph on code speeds was prepared to provide instructors and others with an under- standing of the various bases for computing code speed and with instructions for cutting tapes which would have the exact speed desired. Two studies were made of errors, one of errors in receiving and one of errors in send- ing. In both cases the order of difficulty of the numbers and letters was found to be highly constant for students of different levels of ability. Standardized tests of ability to receive code at different speeds were constructed. The prog- 4 SUMMARY ress of several hundred code students was re- corded and tabulated. These data provide in- formation on average rate of learning and upon variability of rate in learning for stu- dents ranging in level from beginners to men able to receive at 25 groups per minute. 13 PSYCHOLOGICAL PRINCIPLES IN MILITARY TRAINING 1.3.1 Improving Instruction (Chapters 13 and 14) The general operating procedure followed by Applied Psychology Panel projects working on training problems was: 1. In the time available, getting as complete a knowledge as possible of the duties men were being trained to perform. This knowledge was obtained by conducting a job analysis, by tak- ing the course, or through observation and by discussions with experienced personnel. 2. Studying the training procedures in use to determine where and just how improve- ments could be made. These improvements usually involved the application of one or more of the following psychological principles. a. Improve the distribution of practice. b. Secure active participation of the trainee. c. Vary the practice material. d. Develop accurate performance records. e. Give the men an immediate knowledge of the results of their practice. f. Write clear, detailed plans for the in- structor. 3. Putting the improvements into effect. This step sometimes involved the writing of lesson plans. Sometimes it consisted of design- ing a new trainer or improving an existing one. Frequently it included the development of bet- ter methods of measuring the achievement of men in training. 1.3.2 Lesson Plans (Chapter 15) Lesson plans furnish an instructor with a detailed statement of the topics to be covered in a course and the time and method of teach- ing each topic. With inexperienced instructors, lesson plans make better teaching possible and provide a basis for more standardized instruc- tion. Lesson plans were written for many of the special types of training covered in Chapters 2 to 12. The details of writing lesson plans are described in Chapter 15. 1.3.3 The Use and Design of Synthetic Trainers for Military Training (Chapter 16) A good trainer has three essential charac- teristics : practice on it leads to substantial im- provement on the equipment the men are being trained to operate; it provides reliable infor- mation on the quality of the men’s perform- ance ; and, mechanically and electrically, it is as simple as possible. It is necessary to plan carefully an instruc- tional program using a trainer. Simply telling men to practice for a while sometimes results in a loss of skill instead of a gain. Instructions for the proper use of a trainer should be pre- pared as the trainer is being developed, and distributed with the trainer. Trainers possess some advantages over real equipment for practice: they are generally safer, more economical, and more readily avail- able. On a trainer it is possible to break up a complex task into simpler elements. It is usu- ally easier to give men exact information about their errors and successes on a trainer than on real equipment. 1.3.4 Measuring the Effects of Training (Chapter 17) Good tests of the amount of skill acquired by men undergoing training improve the training program in several respects. 1. They increase the amount the trainees learn, both by motivating the trainees and by showing the instructor where his instruction is good and where it is poor. 2. They provide measures of actual achieve- THE DESIGN AND OPERATION OF SPECIAL TYPES OF MILITARY EQUIPMENT 5 ment which are useful in advanced classifica- tion. 3. They provide better criteria than do ordinary school grades for evaluating selection procedures. 4. They make possible more uniform instruc- tion in different classes and different schools. 5. They provide quality control officers with continuous checks on the success of training programs. 1-4 THE DESIGN AND OPERATION OF SPECIAL TYPES OF MILITARY EQUIPMENT Psychological studies of the best methods of operating military equipment, as distinct from the best methods of training, grew to be a more and more significant part of the work of the Applied Psychology Panel as World War II progressed. Studies of operating pro- cedures led in some instances to modifications in design in order to produce equipment better adapted to the abilities and limitations which characterized the average soldier or sailor. 1-4-1 Antiaircraft Directors and Guns (Chapter 18) The accuracy of antiaircraft firing was in need of improvement. Applied Psychology Panel projects studied problems related to the opera- tion and design of antiaircraft equipment. They studied tracking controls; telescopes and sights; operating, maintenance, calibration, and ad- justment procedures; and training features for antiaircraft equipment which determine its ac- curacy. In each case the work was aimed at making operation by the average gunner easier and more accurate. These studies led to a number of specific im- provements in operational procedures and to some design changes. It was demonstrated, for example, that track- ing through large changes in elevation is made easier by gearing the gunsight in such a way that the line of sight of the system will elevate through 90 degrees while the exit pupil of the telescope moves through a smaller (45 to 60 degrees) arc. Field Artillery Equipment (Chapter 19) An analysis of errors in the operation of field artillery equipment led to the conclusion that two types of studies were desirable. The first study analyzed the scales used for panoramic telescopes and resulted in the development of an odometer type of scale which reduced greatly the number of errors made in reading the scale. The second study resulted in the development of remote indicating equipment which recorded the entire action of a field artillery battery. The project was terminated before either study was complete, and before its work could affect actual firing practice. The methods and equipment are available. They should be em- ployed in a continuing study by the field ar- tillery in order to decrease the number of firing errors. 1-4-3 B-29 Gunsights (Chapter 20) An experimental test apparatus was devel- oped for the study of the B-29 gunner in rela- tion to his equipment. This apparatus provided for ground and airborne scoring of perform- ance against synthetic targets. A ground and an airborne synthetic trainer were developed from the experimental apparatus. Experimental studies indicated that trigger- ing the B-29 gunsight occurred semirhyth- mically and independently of the accuracy of fire. Continuous firing was recommended. A set of simplified hand controls for the B-29 gun- sight was developed. They proved to be superior to the standard controls. A study of slewing methods indicated the need for attention to slewing in training and in the design of equip- ment. Viscous damping of the B-29 gunsight was shown to be superior to friction damping. 1-4-4 Radar Equipment (Chapter 21) A number of experimental studies were made of radar operating procedures and of types of oscilloscope presentation. Desirable levels of trace brightness and scope illumination and 6 SUMMARY desirable periods of scope observation were de- termined. Alternative presentations were com- pared in terms of accuracy of determination of target position and range. Procedures for minimizing calibration and operating errors were recommended. It was demonstrated that continued scope operation did not have a harm- ful effect on vision. Stereoscopic Rangefinders and Heightfinders (Chapter 22) Records of stereoscopic rangefinder and heightfinder performance made it apparent that the actual ranging accuracy of the best observers did not approach the predicted, or theoretical, accuracy of the instrument. A num- ber of operating procedures and devices to im- prove performance on the existing equipment were developed. 1. An interpupillometer and template to en- sure that the initial interpupillary setting of the instrument is adequately accurate. 2. Operation of the instrument at reduced aperture whenever possible. 3. An improved calibrating procedure and an improved record form for use in calibration, 4. Special training in making the height-of- image adjustment. Standard Procedures in Voice Communication (Chapter 23) Experimental studies were conducted to aid in the development of courses of instruction in voice communication procedures for AAF per- sonnel. The following facts were established relative to the best methods of speaking and the best ways of using airplane interphone and radio telephone equipment in the presence of intense noise. 1. The most important factor in the use of the voice itself is loudness. In order to secure maximal intelligibility, the speaker should talk in such a way as to produce a good loud side tone in his earphones. 2. The second most important factor is ar- ticulation. One hour of instruction produces enough improvement in articulation to increase intelligibility significantly. 3. Message forms should be standardized. Message content should be standardized. Each type of message should be as unique as possible. 4. The T-17 (hand-held) microphone should be held lightly touching the speaker’s lips and parallel to the plane of the face. 5. The T-30 (throat) microphone should be worn on or slightly above the Adam’s apple, never below it. 6. Gain control should remain inoperative on the interphone. 15 GUIDING PRINCIPLES FOR FUTURE EQUIPMENT DEVELOPMENT i.s.i principles Gf Good Equipment Design (Chapter 24) If an instrument design is well suited to the human operator, it permits many men to qualify as operators, it permits qualified men to operate with efficiency, it permits easy train- ing, and it is acceptable to operators. The general procedure in evaluating the psy- chological efficiency of a design is described. A first attempt at providing a check list for use in evaluating new designs is presented. 1-5 2 The Development of Standard Operating Procedures (Chapter 25) The satisfactory and efficient analysis of operating methods requires the understanding of the military problems involved in the use of the equipment under consideration, in their mechanical, mathematical, psychological, and tactical aspects. The specific steps to be taken in achieving this understanding and developing the operating procedures are the following. 1. Study the equipment. Learn what it is sup- posed to do and how it works. 2. Determine all the tasks which have to be done in adjusting and operating the equipment. 3. Determine what the standards of accuracy of instrument operation should be and how limitations or approximations in the design of the equipment influence these standards. FUTURE RESEARCH ON TRAINING AND EQUIPMENT DESIGN 7 4. Determine how each unit task should be done in order to achieve the greatest efficiency in time and accuracy. 5. Determine the proper sequence of actions. 6. Examine the sequence for short cuts. 7. Try out the procedure or compare alter- native procedures if more than one has been developed. 8. Evaluate operator acceptance of the pro- cedure. 9. Standardize the procedure finally estab- lished. These steps should culminate in the prepara- tion of a manual of standard operating pro- cedures to accompany a new piece of equip- ment as it is distributed for use. 16 FUTURE RESEARCH ON TRAINING AND EQUIPMENT DESIGN The Applied Psychology Panel’s recommen- dations for future research on problems of military psychology are contained in the fore- word to Volume 1 of the Summary Technical Report by this Panel. That foreword also out- lines the type of research organization which the Panel believes most likely to lead to contin- ued successful research in military psychology. Chapter 2 SEARCH-RADAR AND BOMBING-RADAR OPERATORS Donald B. Lindsleyil SUMMARY The training of search-radar and bombing- radar operators is described. Surveys of radar operator training schools led to a number of specific suggestions for improving the in- struction given. The task of the search-radar operator is in general the same, even though there are wide differences in equipment and in special pro- cedures. Common elements in the task include the tactical use of radar equipment, specific procedures such as calibration and tuning, manipulation of controls, interpretation of scope phenomena, and the reading of ranges and bearings. The types of knowledge necessary for each of these frequently occurring tasks are discussed. Several mechanical-optical plan position indi- cator [PPI] trainers were developed to aid in the training of radar operators. A flash-reading method was developed to train men in the rapid identification and interpretation of scope phe- nomena. Studies were conducted of the training of radar bombing operators. An H2X film trainer and radar scope movies were developed for ground training and for use in briefing men for training missions. An experimental study of improvements in bombing accuracy showed that the addition of 150 hours of main scope operating time pro- duced reductions in average bombing error of from 1,000 to 1,500 feet. 21 INTRODUCTION The training of the radar operator can be discussed only in relation to the special tasks he has to perform in connection with the oper- ation of specific kinds of gear. Therefore a particular radar program representing Navy search radar and another representing Army bombing radar will be reviewed. To discuss these programs meaningfully requires the pres- entation of an outline sketch of the training together with specific criticisms and sugges- tions directed toward the weaknesses of the programs. Each program will be followed by general suggestions or the reports of further experi- ments bearing on factors which facilitate training and enhance final proficiency of the radar operator. The emphasis is on search- radar operations. The selection of radar operators is consid- ered in Volume 1, Chapter 6, the Summary Technical Report of the Applied Psychology Panel, and the general questions of the design of radar equipment in Chapter 21. 2 2 SEARCH RADAR 2-2-1 Training the Search-Radar Operator in the Navy The sudden and large demand for radar oper- ators, both in the Army and in the Navy, meant the rapid establishment of radar operator schools, the selection of instructors from rela- tively inexperienced personnel, the securing and setting up of new equipment for demon- stration and training purposes, and the grad- uation of large numbers of operators required to meet quotas. Typical of the emergency train- ing school were two set up by the Navy for training search-radar operators, one at Vir- ginia Beach, Virginia, and another at Point Loma, San Diego, California. About one year after these schools began training operators, the Training Division, Bureau of Personnel, requested Project SC-70, NS-146, Applied Psy- chology Panel, NDRC, to visit these schools and analyze the training programs. The results of these investigations0-8 will be summarized as a basis for further discussion of the problem of search-radar operator training. a This chapter is based on the work of Project SC-70, NS-146. 8 SEARCH RADAR 9 Length of Course and Training Load The length of the course in both schools was 3 weeks, during which time the trainee was expected to become familiar with the operation of several types of radar gear. In addition to learning to operate each type of gear, the operator was required to know something about target detection and interpretation, interfer- ence, jamming, antijamming, CIC, fighter di- rection, navigation, dead-reckoning tracking, plotting, and intercommunication procedures. Obviously only a smattering of knowledge on any of these topics could be acquired in the time available for training. The training load at Virginia Beach at the time of the investigation consisted of 120 men each week, making a total of 360 men in three stages of training at a time. At Point Loma the weekly quota was 200 men, with 600 in training at a time. Only 300 could be handled at a time in the latter school, necessitating two shifts. Analysis of Training Programs In both schools the total training time dur- ing the 3-week period averaged about 105 hours. The percentages of this time devoted to specific aspects of training were as follows: lec- tures, 20 per cent; laboratory, 50 per cent; study periods, 13 per cent; quizzes and reviews, 10 per cent; and instructional films, 7 per cent. For a course 3 weeks in length, this was an acceptable distribution of time. The content of the lectures and laboratory periods was of greater importance. Lectures consisted of radar theory, 55 to 60 per cent; radar usage, 25 to 30 per cent; plotting, 15 to 20 per cent. Laboratory periods were composed of demonstrations and operation of gear, 70 to 75 per cent; plotting, 15 to 25 per cent; trainers and communication, 5 to 10 per cent. Criticisms and Suggestions The following comments were made on course content and laboratory training. 1. The lectures and the operator’s handbook should place more emphasis on practical and functional aspects of operation and operator skills. Many instructors had the misguided no- tion that an operator had to have radar theory and maintenance training, much as a radar me- chanic would be trained. 2. Laboratory and operational procedures should have adequate supervision. Students were assigned to operate equipment with in- sufficient knowledge of correct operating pro- cedures. Thus much valuable time assigned to search watches and operation was wasted by trial-and-error procedures which often resulted in poor habits of operation. 3. Laboratory work should take up one kind of equipment at a time. Operators were shifted from one gear to another before completely mastering any one. This led to confusion and failure to assimilate any specific procedure. 4. Plotting and operating periods should be organized with specific problems in mind rather than relying on rote memory procedures and random operation. The problems should stress tactical and functional uses of the gear. 5. Greater emphasis should be placed upon speed and accuracy in operation. The correct procedure should be taught and the methods of greatest accuracy stressed. When an operator has mastered the procedure, gradual increase in speed of operation should become the goal. 6. An operator should be taught that calibra- tion and tuning are basic to efficient operation and that the greatest care must be observed in carrying out these procedures. 7. Greater recognition should be given to the fact that radar operation is a skill which may be developed to a high degree. Knowledge of the operating procedures is not enough. Skill comes only with practice in the job. Therefore every opportunity should be afforded for the trainee to exercise his knowledge by operating the gear. In all practice he should work for correctness, accuracy, and speed. The timing and posting of operational scores is a good way to foster competition and provide motivation to work harder at the task of improving skill and speed. 8. With better information concerning even- tual assignment of men to ships, based upon production and commissioning schedules, it should be possible to determine the type of equipment for proper training. This would re- duce to a minimum the number of different types of gear upon which an individual operator 10 SEARCH-RADAR AND BOMBING-RADAR OPERATORS would be given training. By concentrating on one or two types much greater proficiency could be attained. 9. Greater use should be made of training aids, especially visual aids. Models, wall dia- grams, illustrations, slides, and movies are in- dispensable for demonstrating the correct procedures or the basic principles behind oper- ating procedures. These should be coordinated with assignments, lectures, and demonstrations. 10. More time and attention should be given to target interpretation and the dynamic as- pects of scope viewing and reading. All con- ceivable types of targets, land masses, fading, and other phenomena which may occur during operation should be illustrated. Operating pro- cedures for all possible exigencies should be practiced. The basic principles of scope inter- pretation, distortion, and interference should be explained not as theory but as practical working concepts. 11. Greater use should be made of radar trainers and similar devices by which operating problems may be presented, specific components of operating procedures may be illustrated and practiced, and quantitative scores on perform- ance may be obtained. 12. More attention should be given to the development of objective measures of profi- ciency. These should emphasize particularly the performance aspects of operation and the solv- ing of “real” operating problems. 13. Fleet operational requirements were con- stantly changing to meet the demands of new tactics. Many small and frequently some major changes in operational procedures were intro- duced in the fleet. In order that the training in the schools might be kept up to date, it was recommended that instructors be selected for a period of sea duty, rotating with men who have had shipborne operational experience. Not only would this system bring the latest operational procedures into the schools, but it would bolster the morale of the trainees. 2-2 2 Functions of the Navy Search-Radar Operator There are two principal types of search radar in the Navy, surface search and air search. The different radars that function in these capacities often vary considerably in appear- ance and physical layout. The scope presenta- tions, the controls, and the operating proce- dures vary according to the specific uses of the equipment. The radar operator’s task in gen- eral is the same, although there may be wide differences in complexity of operation, speed required, and special procedures. It is difficult to specify procedures for any radar set in a hard and fast way since particular tactical situations or maneuvers may call for variations. Thus it is that an operator, whether of surface- search or air-search radar, must be trained in the tactical uses of his equipment as well as in the specific procedures, such as calibration and tuning, manipulation of controls, interpreta- tion of scope phenomena, and the reading of ranges and bearings. The following uses of the equipment will serve to illustrate some of the problems of operation which depend upon specific training and experience. Type of Search Function Surface-search radar requires that a con- stant 360-degree search be maintained, unless radars in other ships of the formation are assigned specific sectors. The search function is further subdivided into long-range search and short-range search. The former is for the purpose of picking up large surface targets and requires the use of appropriate range scales and proper adjustment of gain controls. In the case of distant surface targets, speed is not so essential as for close-range targets or air tar- gets, and therefore more liberty may be taken in connection with the speed of antenna rota- tion, stopping of antenna rotation for study of target composition, and tracking in order to determine relative motion. The operator should remember, however, the need to make frequent 360-degree scans for possible targets in other sectors. With bearing and range determined accurately for a target, continuous search may be resumed, but a plot of the target should be maintained. An operator should be skillful enough to read ranges and bearings without stopping the rotation of the antenna. This is an aspect of performance which is often not stressed sufficiently in training. SEARCH RADAR 11 Short-range surface search is for the purpose of detecting submarines and small craft or, in the case of offshore operations, for the detec- tion of possible coral reefs or other obstacles to navigation. Sometimes the short range is used for keeping a ship in position in a formation, especially during severe maneuvers. Because the antenna must be tilted down during short- range search, the problem of balancing sea return against possible targets requires con- stant manipulation of gain controls. The use of surface-search radar for navigation along coastlines demands not only a thorough knowl- edge of scope interpretation and familiarity with the geographical features of the area but are difficult to teach in training but which can often be made vivid and real by means of train- ers which produce synthetic targets on the actual gear and which allow not only maneu- vering of targets but also make provisions for scoring the performance of the operator. A PPI flash-reading radar trainer is shown in Figure 1, The trainer was used to teach stu- dents to read the bearing and range of blips quickly and accurately. The trainer consists of a master timing unit and five repeater scopes. Targets may be presented in any desired order and position and with varying persistence.12’1(5 A mechanical PPI tracking trainer17 is shown in Figures 2 and 3. Figure 2 shows the simu- lated scope face; Figure 3 the mechanical con- trols of the target trace. Two flashlight-like projector tubes ride on range cams and are geared to bearing cams. Light from the pro- jectors is focused on a persistence screen and is interrupted by a rotating sector disk. The continuous movement of the projectors and the interruption of the lights by the sector disk simulate the PPI presentation of two aerial tar- gets. The courses may be altered by changing cams. Another teaching possibility is the presenta- tion of tactical problems by means of a series of successive pencil-and-paper test diagrams which may be used as proficiency measures. A surprising degree of dynamic simulation, as well as difficult problems requiring the operator to think through the solution, may be worked out in this way. Examples of this kind of func- tional and tactical problem may be found in tests,9’19 workbooks,10’11 and manuals13 devel- oped by Project SC-70, NS-146, for the Services. Operating air-search radar is usually more difficult than operating sea-search radar. The greater difficulty is due to the speed of search and tracking and the complexity of identifying and keeping a plot of friendly and enemy planes. Long-range air search is, of course, for the purpose of detecting and identifying enemy planes at maximum range. The nature of the pip or blip must be studied carefully in order to determine the number, type, and movement, if possible, of the attacking force. Interceptors, if available, must be alerted and vectored to the attacking force. Throughout these activities the Figure 1. Close-up of repeater unit of PPI flash- reading radar trainer, showing illuminated center splash and target complex consisting of fine blips. (The bearing scale in this photograph is in- verted.) an understanding of the characteristics of the radar, such as beam width and pulse length, which contribute to distortion in the mapping of land areas. In short-range operations the operator should be especially careful not to overlook the matter of constant search for pos- sible attack from another quarter. Having de- tected a target at short range, it may be neces- sary to turn it over immediately to fire-control radar and continue full search operations. These are tactical operating problems which 12 SEARCH-RADAR AND BOMBING-RADAR OPERATORS operator must remain alert to the possibility of attack from other quarters, which means that scanning of the entire area must continue. In short-range air search, speed is even more im- portant. Rotation of the antenna must be more rapid, target location and identification must be made more quickly, and reports must be com- municated continuously. From the foregoing brief account of the oper- ator’s task in surface-search and air-search proper operational and tactical procedures were taught in the schools. Knowledge of proper procedures depends upon close liaison between the fleet and the schools; the introduction of these methods in training requires careful plan- ning of the school program so that tactical procedures will dovetail with basic operations of the equipment. In general, more considera- tion should be given to the step-by-step advance- ment through the various stages of training. Whenever possible, those phases of operation which require more practice and skill than others may be separated, as components from the whole task, and given special attention. Calibration and Tuning Calibration and tuning can usually be ac- complished by two or three alternative meth- ods, and since an operator may be forced to use any one of them he should know them all per- fectly. Not only should the operator know the methods; he should have opportunity to prac- tice them so that all the steps become automatic and habitual. If opportunity to see the alterna- tive tuning methods function is limited with land-based equipment, an attempt should be made to provide synthetic signals simulating the kinds of returns the operator will be forced to use. Emphasis should be placed upon the fact that an adequate warm-up period for the equipment must be allowed before calibration and tuning may be accomplished accurately. Knowledge of and Use of Controls The operator not only must know the func- tions and limitations of all controls but must be able to manipulate them skillfully and rapidly. The discrete function of each control is only part of the total function, and much must be learned by experience with regard to the refine- ments which may be accomplished by delicate adjustments of two or more controls simul- taneously. For example, it is sometimes neces- sary to attain a delicate balance by manipulat- ing intensity, focus, and receiver gain controls in order to bring out a target echo clearly. The operator should know the exact location of all controls so that he can reach for them without removing his gaze from the radar scope. This type of tactual-kinesthetic adjustment is ac- Figure 2, Mechanical PPI tracking trainer, scope face. operations it may readily be seen that the com- plexity of the task depends upon the situation. Much more is required of training than that it teach the fundamentals of operation. Although the elements of operation are essential, there must be superimposed upon and integrated with that kind of training the complete gamut of tactical operations. The latter will of course be- come refined and most highly efficient only with extensive experience in the fleet, but it is essen- tial that practice in such operations begins early and is integrated with elementary oper- ator training. This is important in order to avoid the bad habits which invariably become associated with operator training when it is not properly oriented toward the final goal of achievement. Too frequently it was reported from operational sources that habits of opera- tion learned in basic training had to be un- learned and new ones substituted. This should not have been necessary if from the start the SEARCH RADAR 13 quired only with extensive practice with the actual gear. Even though an operator has access to the gear, frequently he may have little oppor- tunity to make the required adjustments be- cause of lack of target situations, unless some type of synthetic signal generator (radar his position, course, speed, type, and strength, if composed of more than one ship or plane. His position is obtained by range and bearing. His course and speed may be determined by plotting a series of positions as a function of time. His type and strength, as well as other Figure 3. Mechanical PPI tracking trainer, controls trainer) is available for feeding targets in under a variety of conditions. Target Identification and Interpretation Less than half of an operator’s knowledge and skill are complete when he has mastered start-stop procedures, calibration and tuning, manipulation of controls, and the finding of the bearing and range of a target. Screen inter- pretation and the ability to analyze an echo in terms of its composition are the critical aspects of successful operation. First of all a target echo must be recognized as friend or enemy. This involves knowledge of information friend or foe [IFF] procedures and the recog- nition and interpretation of the coded signals which appear on the scope and are the basis for such recognition. If the contact is established as an enemy, the next step is to secure as much information as possible about him, including possible characteristics, may be determined by piecing together a number of bits of informa- tion, some of which are dependent upon knowl- edge of set characteristics, some upon range and bearing information, but principally from the size, form, and dynamic aspects of the target echo on the screen. For example, the nature and amount of bobbing or fluctuation of the pip on an A scope is sometimes a cue to the distinguishing of a single or multiple plane target. A single pip may sometimes show by its breadth and by multiple peaks that a con- tact is composed of a large number of planes. To recognize a single blip on a PPI scope as a single large target, or as two or more small targets, depends upon knowledge of range and bearing resolution, distortion due to pulse length and beam width, and other factors. Although the theory of target analysis may be taught, the use of such knowledge is depend- 14 SEARCH-RADAR AND BOMBING-RADAR OPERATORS ent upon actual practice with a variety of target conditions. These can seldom be supplied in a land-based school due to the sparsity of targets and lack of variety of target situations. In order to make the time during which an operator is practicing on the actual gear maximally useful, a radar trainer or other simulating device for feeding in synthetic signals should be used. Such trainers must provide good simulation of the types of target conditions which are needed for practice in target identification. If electronic signal generators are not available for this purpose, some simpler type of mechanical- optical12’ 15’17 device should be developed for presenting the target situations on a mock-up scope. In fact this type of device might well supplement the use of the electronic simulator. Another way in which practice in the study and interpretation of target composition may be made possible is through the use of a series of photographs, or better, a movie18 of actual scopes during the kinds of target situations which it is desired to study. Such pictures may be taken on board ship during actual missions or during experimental cruises. The still pic- tures may be arranged in a problem series re- quiring the trainee to work out a solution. The movies may be presented on a mock-up radar scope with provision for reversal and detailed study of characteristics. 2'2'3 Trainers and Training Situations Three studies4’3> 7 have shown that electronic trainers can be used effectively to develop pro- ficiency in certain aspects of a radar operator’s task. This is especially true if a means of meas- uring performance is associated with the trainer so that the student’s score can be made available immediately. Learning curves may be plotted for individuals and for the group of trainees and in this way the optimal period of time to spend in practice on a particular task may be determined. In all three of the above studies, typical learning curves were found. These showed progressive improvement with practice up to a point where the curves leveled off, indicating that maximal efficiency in that function had been attained. One of the major weaknesses found among Navy search-radar operators was the inability to read target locations (range and bearing) rapidly and accurately. In order to provide practice in this task, a mechanical-optical trainer1- was developed, together with a flash- reading method of training. Tne trainer con- sisted of a timing unit which controlled the timing and location of a series of blips which appeared on the mock-up scopes of several scope-reading positions (see Figure 1). Multi- ple or single blips could be presented at varying rates of speed. The position of each blip was known so that the accuracy of the student’s readings could be immediately checked and his results posted for comparison with the speed and accuracy attained by other students. This not only provided a measure of proficiency by which an instructor could gauge progress but it served as a motivating device to encourage students to work harder to improve their skill in reading range and bearing of target echoes. Trainers of this type were used successfully at the Virginia Beach and Fort Lauderdale train- ing schools, where they were coupled with reporting procedures and the use of sound- powered phones. During the training period, records of battle noises were played in order to provide some degree of operational reality. The method of training and some preliminary results obtained at Fort Lauderdale have been described.16 An important feature of this method of training was that simple targets and a slow rate of presentation were given initially until an operator had the habit of reading blip position accurately; then the rate of speed was stepped up and the number and complexity of the echoes were increased. It should be strongly emphasized that prac- tice on a trainer which does not provide a measure of performance so that the student can appreciate his progress is likely to be a waste of time. In one study1 where knowledge of results was not made available to the student, it was demonstrated that accuracy of performance got worse instead of better with increased prac- tice. Training by means of trainers must be closely supervised to ensure that students de- velop the correct habits of operation; otherwise the training may prove detrimental. RADAR BOMBING 15 22 4 Final Achievement Examinations for Search-Radar Operators As mentioned previously in connection with the survey of training programs in search-radar operator schools, there was found to be a defi- nite need for improvement of the examination methods. In particular, it was deemed advisable to develop objective-type examinations as meas- ures of final achievement for use at the end of the training period. As a result of a conference called by the Bureau of Naval Personnel which included representatives from the three main- land operator schools and the Fleet School, Pearl Harbor, it was recommended that Project SC-70, NS-146 prepare a series of final achieve- ment examinations for use in the mainland schools. This study14 required extensive work with each school and the preparation of objective examinations covering the operation of a num- ber of different types of radar gear. Nineteen experimental tests were made and tried out in the schools. An item analysis was made to rule out items which did not contribute significantly to the total examination. Finally, three radar operator final achievement batteries20 of two forms each were prepared and submitted to the Bureau of Naval Personnel for reproduction and use in the schools. Each examination in- cluded sections on basic radar, two or more types of radar gear, DRT plotting, relative mo- tion, navigation, air plotting, and surface plot- ting. 2 4 RADAR BOMBING 2-31 Army Training of Radar Bombing Operators The second type of radar training program was for bombers. Initially, enlisted men were trained as airborne radar operators. Later, when navigational and bombing procedures be- came intimately associated with the use of radar, navigators and bombardiers were trained to operate the radar equipment. For the most part the training of the radar observer (bombardment) [ROB] or the training of the radar operator for high-altitude radar bombing- allowed adequate time and provided a well- balanced course. There were, from time to time, changes in the length and content of the course, but for the most part these changes were consistent with improvement of the course of training and in line with recommendations from the theater of operations. The principal difficulties encountered in training were in ad- justing training schedules and quotas to new demands made by higher headquarters. There were constant shifts in the plans as to who would be trained, what equipment they would be trained on, and what operation procedures would be followed. These variables made it al- most impossible to standardize the course of training and to apply the necessary proficiency measures, although despite the constant changes considerable progress was made in these areas. Briefly, the course of training, which was 10 weeks in length, consisted of radar ground training, 360 hours; flying training, 80 hours; and miscellaneous training, 85 hours. Ground training for ROB consisted of navigation re- view ; bombing review; basic radar theory; fundamentals of one specific type of radar bombing equipment, including component parts, calibration and tuning, controls, and trouble shooting; accessory equipment; radar naviga- tion; radar intelligence, consisting primarily of scope interpretation and target study; radar bombing methods and procedures; briefing and critiques; and supersonic trainers. Flying train- ing required at least 35 hours of operation of the main scope and 30 hours of observation of an auxiliary scope. Miscellaneous training in- cluded instruction in the use of personal equip- ment, such as parachutes, oxygen masks, oxy- gen equipment, and emergency equipment; military training; physical training; war orien- tation ; and chemical warfare training. Deserving of special comment is the fact that radar theory was rightfully given a relatively small amount of time and that scope interpre- tation and target analysis were given an ade- quate proportion of time. A large amount of detailed procedure was necessary under radar bombing and radar navigation, but, since these procedures all came in for varying amounts of practice during flight training and in the use 16 SEARCH-RADAR AND BOMBING-RADAR OPERATORS of supersonic trainers, it appears that the amount of time assigned to them in lectures and demonstrations was adequate. The most out- standing weaknesses of the program were in briefing procedures, use of supersonic trainers, and scope interpretation. The time devoted to briefing and critiques of missions was minimal but probably adequate. Pre-mission briefing frequently consisted of weather reports, a hasty survey of the route of the mission, and a brief discussion of some possible difficulties to be encountered in navigating to the target area. A great deal more could have been accom- olished by way of showing movies and stop- frame pictures of the radar scope taken espe- cially for briefing of the route to be followed, so that the radar scope navigation problems the development of specific operational skills. This could have been remedied by careful and complete briefing. The H2X film trainer15 and the radar scope movies18 developed by Project SC-70, NS-146 were prepared with the facilitation of the briefing task in mind, al- though the trainer and the films could also be used to good advantage in scope interpretation studies. The H2X film trainer15 is shown in Figures 4 and 5. Figure 4 shows the front view of a photographic mock-up of an AN/APS-15 re- ceiver-indicator unit. A repeater projector, mounted as shown in Figure 5, projects 16-mm movies by means of a mirror system upon the rear projection screen in the scope opening. Scope movies taken on actual navigation and bombing missions are projected. The realistic effect is useful in familiarizing the student with the scope presentation as seen in the air, in teaching scope interpretation, and in briefing men for practice missions. 2 3-2 Phase Checks and Final Proficiency Measures Because of the length of the period of train- ing and the diversity, there was a serious need for systematic evaluation of progress by phases. Also as in all other training programs, there was a need for some means of final evaluation of proficiency in order that an operator might be classified and assigned, as well as that the school might know when an operator had reached a given level of proficiency. Without some valid measure of proficiency it was im- possible for the school to set a criterion of final proficiency and thus scale its training program to meet this requirement. In the summer of 1943, Project SC-70, NS-146 began work on a group of final profi- ciency measures for operators of air surface vessel [ASV] radar. These measures2 and the method5 of construction have been described. Later with the introduction of radar bombing equipment other measures of proficiency were constructed19 for specific equipments. In the fall of 1944, the Air Surgeon’s Office assigned a group of men from the Psychological Research Figure 4. H2X film trainer, front view. could have been studied and digested in rela- tion to the maps. A preliminary study of scope movies or photographs of the target area would have greatly improved the efficiency of the training on bombing runs. In navigating, find- ing the initial point, and finding the aiming point, the radar operator frequently became confused due to unfamiliarity with the region and the type of scope returns. Although the ability of an operator to find his way on blind missions was important and should have been emphasized toward the end of operational training, the uncertainties of navigation and target identification should not have been al- lowed to become a distraction detrimental to RADAR BOMDING 17 point bombsight range error, ground-speed range error, and time-of-fall range error. The experiment showed that the amount of additional training required to reach peak effi- ciency was 85 hours, over and above that regu- larly given, when the total group of operators was considered. It was interesting to note that the navigator-trained operators required only 35 hours of additional training to reach peak Division to work specifically on the develop- ment of proficiency measures for radar bomb- ing equipment. This group, known as the Psy- chological Research Project (Radar), did an excellent job in attempting to provide system- atic phase checks and final proficiency measures for the constantly changing radar bombing program. Their work was instrumental in help- ing to standardize the training program in the various schools. 2'3'3 An Experiment on Training and the Analysis of Radar Bombing Errors When radar observers (bombardment) began to reach the active theaters there were numerous complaints that the accuracy of bombing by radar was not satisfactory. It was not known whether this was due to the type of equipment being used, to inadequacy of train- ing, or to insufficient length of operational fly- ing training. In order to solve this problem the AAF Training Command, Fort Worth, decided to set up an extended training experiment at the AAF Training School at Victorville, Cali- fornia. Project SC-70, NS-146 was asked to supervise and report upon the results of this experiment.21 The study was carried out during the period from April to July 1945. During this time 20 graduates of the regular course in radar bom- bardment, including 10 bombardier-trained operators and 10 navigator-trained operators, were given 150 hours of main scope operating time. This was in addition to the 35 hours of main scope time during the regular training course. Since the method of synchronous bomb- ing was used throughout, 20 experienced bom- bardiers returned from combat were assigned to operate the bombsights. Each bombardier was teamed with a radar operator for the dura- tion of the experiment. Eleven different target areas were used throughout the training and at intervals a four-target test mission was flown. Bombing proficiency was scored by the photo- bomb-scoring method and a detailed analysis of the sources of error was made. The components of error for which analysis was made included drift error, final point deflection error, radar altitude range error, radar range error, final Figure 5. H2X film trainer, projector mounting. efficiency, thus raising the question concerning the selection of men to be trained as radar observer (bombardment). The analysis of type of error showed that the two largest sources of error were the final point deflection error and the radar range error. Both of these indicate inaccuracy in localizing the precise aiming point during the bombing run. This result shows that scope interpretation was largely at fault and that considerably more em- phasis should be placed upon training in target analysis and scope interpretation. The third most important source of error was the radar 18 SEARCH-RADAR AND BOMBING-RADAR OPERATORS altitude range error, much of which was due to improper calibration. Thus emphasis should also be placed upon set calibration. During the course of the additional training there was a consistent reduction in the circular error, range error, and deflection error. These reductions from the first 25 hours to the last 25 hours of the training period were approxi- mately 1,500 feet, 1,000 feet, and 1,000 feet, respectively. The results of the experiment pro- vided information on the general level of bomb- ing accuracy which could be expected with the particular kind of equipment used and clearly indicated that with additional training of the operational type much greater proficiency could be attained. Chapter 3 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROF William C. Biel and William E. Kappauf, Jr.a SUMMARY The problems found in the training of track- ers include those of obtaining objective measures of the tracker’s skill, of his learning speed, and of the nature of his learning task. Since the tracker uses complicated, expensive equipment, the problem of designing supple- mentary trainers is important. Objective measures of the tracker’s skill are more reliable and useful than instructor’s rat- ings; ratings are unreliable and do not agree with accurate methods of scoring performance. Checksights, found to be the most generally useful objective measure, were developed for use on various guns. Checksight scores may be obtained by observing whether the tracker is on target at certain instants, such as once every 2 seconds, or observing the tracker at pre- selected bearing points. A phototube scoring device to replace the checksight observer was developed too late in the war to be used. Its reliability and effectiveness were, however, demonstrated. Trackers in general learn more rapidly when trained with checksights than when checksights are not used. The learning curves demonstrate the time when improvement ceases and thus make it possible to judge when the training program should end. Standards for daily improvement can be set up for each tracking device and the conditions for its use. These standards serve as a guide for both instructors and trainees. Learning records show that, once men are adequately trained in tracking, they need refresher drills no more often than once a month. Three on-target tracking trainers (trainer means a supplementary device for use in train- ing) were developed: a trainer which simulated the task of optical tracking with the stereo- scopic heightfinder; a trainer which simulated the two-man tracking task in the gun director Mark 37; and a trainer which simulated the two-man tracking task on the director M7. The last one became known as the Tufts tracking trainer. It has validity as both a selection device and a training device. Several manuals giving instructions for the use of gunnery trainers and the operations of fire-control equipment were written. 31 VARIETY IN TRACKING TASKS The task of tracking, broadly defined, may be stated as the operation of advancing the posi- tion of a sighting system or a gun so that it continually points in a direction which bears some required relation to the position of the target. Some of the complication of this defini- tion arises from the fact that the name “track- ing” is applied to two similar operating jobs in which the point of aim is different. These are on-target tracking and lead tracking. In on-target tracking, the tracker keeps a line of sight on the target: he keeps the target as near as possible to the center of the reticle or ring sight which establishes the line of sight. In lead tracking, which is used only by gun- ners and only in some modes of gun opera- tion, the tracker keeps the target off center in the reticle or sight; he keeps the target dis- placed in a direction and to an amount which he estimates to be necessary for the gun to aim at the place where the target is going to be by the time the projectiles reach it. Thus, lead tracking involves lead determination. On- target tracking, on the other hand, is tracking per se, and is used when leads are established by auxiliary fire control or computing systems. In on-target tracking, the tracker’s function is to establish the train and/or elevation posi- tion of the target relative to the battery and the rates at which that position is changing. Most of the wartime research work on the training of trackers concerned on-target train- ing. The tracking job on any particular piece of equipment is not completely specified until it a This chapter is based primarily on the work of the Applied Psychology Panel projects listed in the chapter. 19 20 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROL is further described in terms of the mechanics, perceptual tasks, and controls involved. For purposes of orientation, a review of these de- scriptive terms and tracking variables follows. Tracking may be classified as direct line-of- sight tracking or disturbed line-of-sight track- ing. In the former, the line of sight of the tracking system is fixed relative to the sight or telescope mount. The line of sight always moves with the mount and through the same angles as the mount is moved. In disturbed line-of- sight tracking, the line of sight of the tracking system is offset relative to the mount axes by a computing system which functions to deter- mine the angular leads for firing. When the line of sight is on the target, the mount is aimed in the direction in which the guns should fire. Motion of the mount does not produce an identical motion of the line of sight, because the position of the line of sight is affected not only by the tracking movements but also by the action of the computer. Another classification of tracking contrasts optical tracking and radar tracking. The track- ing function is the same in both cases. Either the optical line of sight or the radar line of sight is kept on target. The main difference between the two forms of operation is found in the kind of visual presentation which the tracker views. Distinction is also made between the types of tracking control. Tracking may be by direct control, as when handle bars or handwheels are attached or geared directly to the tracking structure. This is sometimes called position tracking. It is contrasted with the form of tracking known as rate tracking or velocity tracking. In this case, the tracker adjusts the speed of a power drive and thereby regulates the rate of motion of the tracking structure. Both these forms are different from rate con- trol, also known as aided tracking. In this case, the tracking movement controls the position of the line of sight at the moment but also varies the rate of motion of the line of sight as controlled by a power drive. Another set of tracking variables, which runs through most of the above classifications, con- cerns the specific form of the control mechan- ism which the tracker uses, be it handwheel, handle bar, handle grip, joy stick, or the like. Any of these controls may be adapted to single or to two-man tracking situations. The variety in tracking tasks indicated by this discussion makes it apparent that no war- time research program on training methods could hope to examine the specific training problems met within each set of tracking con- ditions. Only selected training problems were dealt with. Those which were investigated by the Applied Psychology Panel, in particular, were for the most part problems proposed by the Services. The Panel cooperated by under- taking direct research in training methods, by developing training instruments and methods of using them, and by preparing training ma- terials, lesson plans, and the like. Contributing to this work were the following projects: the Height Finder Project, N-105, N-lll, N-114, NS-146, SC-70, SOS-6, and AC-94. In the last- named project the tracking task was compli- cated by the fact that the personnel not only tracked but simultaneously ranged on a target. The work of Project AC-94 is described in Chapters 5 and 20. In this chapter will be con- sidered the research on tasks involving tracking only. 32 MEASURING TRACKING SKILL Methods of scoring trackers attract the in- terest of the instructors for at least three rea- sons. First, they want to know when their men are adequately trained. Second, they want to know which men in a group are the best track- ers. Further, if they have had some introduc- tion to the psychology of learning, they want to provide their men with tracking scores during their training to improve their rate of learning and their final level of tracking proficiency. The paragraphs which follow discuss and evaluate various methods, devices, and schemes for scoring trackers or measuring tracking skill. 3-21 Inadequacy of Qualitative Observa- tional Methods Many gunnery and fire-control instructors accept as obvious their ability to rate trackers MEASURING TRACKING SKILL 21 and to know when a man is tracking well. They judge tracking skill on the basis of how a man grasps the tracking handwheel, how he turns the handwheel, or how smoothly he moves the director or the gun. Attractive as these qualita- tive scoring methods may seem, experimental evaluation indicates that they have little prac- tical usefulness. A study by Project SOS-6 of the Applied Psychology Panel demonstrated this by comparing judgments of tracking made by qualified antiaircraft officers with actual checksighting scores.23 Three officers observed and scored the track- ing of 30 trackers. Each of the men was scored on two courses in azimuth and two courses in elevation. The men were rotated in such a way that when a man had tracked two courses on one side of the gun, he did not track his two courses on the other side of the gun until some eight courses later. This minimized the possibility that the judges would generalize from azimuth to elevation ratings and vice versa. The target was an AT-11, flying a crossing course, average speed 160 mph, altitude about 1,000 feet, mini- mum range about 1,200 yards. The three officer observers were instructed to watch the tracking as carefully as possible and to use all the tricks or cues they knew in evaluating the tracking. For each pair of courses by a given tracker, each judge was to record the course on which the tracking had been the better. At the end of the test, each judge was to rate the men in order of tracking ability, using, in his ratings, notes taken during the tracking runs. All judgments were made separately for azimuth and elevation tracking. The scores and ratings provided by the offi- cers were compared with the checksight scores obtained for the same courses and periods of tracking. The method of checksighting used was that of observing whether the tracker was on target at certain times; the method is de- scribed in Section 3.2.4, The results were these: of the 30 pairs of courses tracked in azimuth by the 30 men, 28 were such that the checksight scores on the first and second runs of the pair were different. These 28 paired courses, judged by the three officers, yielded a total of 84 judged compari- sons. Of these, 48.8 per cent were in agreement with the checksight score differences and 51.2 per cent were in disagreement. In elevation, 24 paired courses yielded checksight score differ- ences. These netted 72 judged comparisons, of which 45.8 per cent were correct and 54.2 per cent were incorrect. Thus the officer ob- servers were just as often wrong as they were right in deciding which of two immediately successive courses represented the better track- ing. When the rankings of the 30 men as deter- mined by the checksight scores were correlated with the rankings provided by each of the offi- cers, none of the correlations was reliably different from zero. The judges had only 25 per cent success in picking out the worst or the best man of the group and only about 31 per cent success in picking out the men in the top or bottom 10 per cent of the group. The generalization of these results for 40 mm gun tracking is that qualitative methods of assessing tracking are not accurate. True, some of the rating factors used by the officers might have been employed more systematically if in- corporated in a carefully developed check list, but other similar studies show that these meth- ods, when perfected, will not be as accurate as objective and quantitative checksight methods. Recording Cameras Recording cameras, which have seen wide use in gunnery and director tests and on aerial combat missions, provide very precise records of tracking. Depending on the degree of detail wished for in the analysis, the cameras may be operated at many frames per second or con- trolled to take one photograph every second or two. Two Applied Psychology Panel projects were interested in the development and use of recording cameras for analyzing tracking with lead-computing gunsights. Project N-105 worked on a camera for the gunsight Mark 14. It is shown in Figure I.29 The gunsight Mark 14 has a disturbed line of sight and uses an illuminated, reflected reticle. In order to photograph both target and reticle, Project N-105 mounted the camera on top of the gunsight where it was directed down at the 22 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROL elevation mirror in the sight. With this ar- rangement, the reticle was not disturbed in elevation, as far as the camera was concerned, but it was disturbed in traverse. Since the tar- get and reticle stayed within the camera field only when the traverse lead stayed below 5 perimental analysis or for instruction during the final stages of the training of director teams. They can hardly be considered as day- by-day training devices, however, because scor- ing of the tracking performance from the films takes too long. For this reason, whenever con- ditions permit, checksights are used instead of cameras for scoring trackers during training. Checksights For use on the 40 mm gun, Project SOS-6 devised two forms of checksight, one made from the 4-power elbow telescope M1A111 and another less elaborate one called the Plexiglas checksight.24 Either of these sights can be attached temporarily to the automatic loader of the gun or permanently to the telescope of the computing sight M7. Both sights have demonstrated value when used for scoring the performance of men undergoing training, but the Plexiglas checksight is less accurate and of lower reliability than the elbow telescope checksight. Figure 1. Gunsight Mark 14 with camera at- tached, front view. degrees, tracking errors could be scored from the camera film only when the traverse lead was less than 5 degrees. This was a limitation in the camera design, although one which was not serious for incoming courses. Improvement would appear to depend upon an extension of the effective field of the camera or the elimina- tion of the disturbed traverse condition by the use of a compensating traverse mirror in the recording system. Project N-lll assisted the Bureau of Ord- nance in the development of a recording camera for use with the gunsight Mark 15. This camera is mounted so that it aims into a semi- reflecting prism on the eyepiece of the gun- sight. It photographs the fixed telescopic reticle in the gunsight, the image of the target, and when used on the gun director Mark 63, the radar tracking dot. The installation is known as test equipment Mark 1, Model 1. These cameras are of great usefulness in recording tracking and firing results for ex- Figure 2. Checksight made from elbow telescope M1A1 attached by bracket to elevation telescope of computing sight M7 on a 40 mm gun. Figure 2 shows the elbow telescope M1A1 mounted on the elevation telescope of the computing sight M7. Figure 3 shows the check- sight in use. The reticle pattern which the MEASURING TRACKING SKILL 23 checksight operator sees in the telescope is shown in Figure 4. Graduations are in 5-mil steps. The central open area extends 2 mils from center in all directions. able on a wide variety of tracking devices. Figure 6 shows it in use on an M55 turret. It can be used for evaluating either on-target tracking or lead tracking. For use on the gun director Mark 37, Project N-114 installed a telescope Mark 79 on the left end of the rangefinder tube about 18 inches outside the director housing. The telescope was collimated with the director. For purposes of Figure 3. Elevation telescope bracket and elbow telescope M1A1 mounted on the M7 telescope on the elevation side of the gun. If a check on track- ing during firing is desired, this arrangement is recommended. Figure 5. Plexiglas checksight. scoring tracking performance, either optical or radar, the reticle in the standard Mark 79 was replaced by one graduated in 2- and in 5-mil steps (see Figure 7).12 For use with the gunsight Mark 15, Project N-lll assisted the Bureau of Ordnance in de- veloping and testing a checksight, now known as test equipment Mark 1, Model 0.33’38 The checksight equipment includes a special reflect- ing prism and mount which clamps over the eyepiece lens of the gunsight; a bracket which attaches to the director; and a unit-power tele- scope which mounts on the bracket and is Figure 4. Reticle pattern in elbow telescope M1A1. Reticle numbers indicate mils. The Plexiglas checksight is shown in dia- gram in Figure 5. It was developed for the special purpose of having a device which could be made in the field and which would be applic- 24 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROL directed at the prism. The telescope is attached to the director as shown in Figure 8 so that anyone looking through the telescope can see exactly what the tracker sees as he tracks. The reticle pattern in the checksight telescope must be aligned with the gunsight reticle. It includes crosslines and a series of circles as shown in Figure 9. The center open area of the reticle represents a circle which is 3V2 mils in diameter in terms of the gunsight field. An interesting use of this checksight for the gunsight Mark 15 is its use in checking the Figure 8. Test equipment Mark 1, Model 0. accuracy of blind tracking with the gunsight on the gun director Mark 63. On clear days, the radar dot and the visible target are nor- mally seen together in the gunsight field, but by the use of mutually exclusive red and green Figure 6. A checksight observer using a Plexi- glas checksight to observe tracking on the M55 turret. Figure 9. Reticle in telescope shown in Figure 8. filters (red over the front window of the gun- sight and green in front of the tracker's eye) the tracker is allowed a view of the radar dot only, while the checksight operator can see the target itself and observe how well the tracker is following it. Figure 7. Reticle pattern in telescope Mark 79 used as a checksight on the gun director Mark 37. The scale is calibrated in 2-mil steps from the center to 10 mils, then in 5-mil steps to the edge of the field. MEASURING TRACKING SKILL 25 Checksight Scoring Methods There are a variety of ways in which the reticle shown in the accompanying figures may be used in scoring tracking performance by checksight observation. These methods require a timing device or a performance sampling scheme of some sort. Projects under the Ap- plied Psychology Panel collected data by five different methods. They are the following. Per Cent of Time on Target This method involves a continuous time-clock or stop-watch record of the per cent of time that a given point of the target (the aiming point) is within a specified distance of the center of the reticle. Two clocks are needed, one to measure the duration of the course, the other to be operated by the checksight observer whenever the tracker is on target according to the agreed-upon criterion. Project SOS-6 used this method of scoring with the elbow telescope M1A1 as a checksight.11 For scoring 40 mm gun tracking with the computing sight M7, the method was found to deliver scores on indi- vidual courses which correlated +.92 with similar per cent time-on-target scores obtained from camera records of the tracking for those courses. When two observers scored the same course using two collimated checksights mounted on the same gun, the correlation of the scores which they obtained was +.84. Proj- ect N-lll tested the precision and accuracy of this same scoring method by having a group of four observers score a series of 30-second tracking runs presented by motion picture. It was found, in agreement with the results from Project SOS-6, that the scores obtained by an individual course differed from frame-to-frame motion picture scoring by only 8 per cent on the average. Both the reliability and the ac- curacy of these checksight scores are, of course, improved if scores are obtained and averaged for a number of courses. Thus, if a tracker is scored for five short courses, the data indicate that the average checksight score obtained for him would be in error, as compared with photo- graphic frame-by-frame scoring, by approxi- mately 4 per cent.10 Per Cent of Observation Instants when on Target This method is essentially the same as the one just described except that it does not re- quire the use of a continuous timing mechan- ism. The checksight operator observes whether the tracker is on target at certain instants, perhaps once per second or once every 2 sec- onds. He receives observing signals from an assistant and merely counts the number of times that the tracker is on target when the signal comes. At the end of the course, the tally is expressed as a per cent of the total number of signals or observing instants which were called. This method is just as accurate as the continuous timing method and is prob- ably more reliable.10 Estimated (Not Measured) Time on Target By this method, a checksight operator merely estimates the per cent of the time that the tracker is on target. How closely he tries to estimate is optional, but in one study by Project SOS-6 the estimate was made to the nearest 25 per cent, i.e., scores were given as 0, 25 per cent, 50 per cent, 75 per cent, or 100 per cent for each course. Correlation of these scores with scores obtained by continuous timing (method 1 above) was +.79 when obtained by a group of three observers for a group of 25 trackers of varying ability. Reliability of the method for a group of 24 courses tracked by one man was about +.50, a value which would probably have been higher if the variation in tracking of this one subject from course to course had been as great as the variation be- tween different trackers.19 A follow-up study indicated that the method is quite satisfactory for use with the Plexiglas checksight in the rough assessment of tracking performance on short-range crossing courses.20 In such courses trackers vary widely. Average Tracking Error A measure of average tracking error is im- possible to obtain by direct visual observation unless the error is estimated at specific inter- vals. Such spaced estimates, one every 10 sec- onds, were made in a series of checksight scoring experiments by Project N-114.27 Er- 26 TRAINING TRACKERS FOR ANTIAIRCRAFT FIFE CONTROL rors, observed in elevation and in train sepa- rately, were estimated to the nearest mil. When checksight operator estimates were compared with photographic records of the errors, the average error of a single estimate was about 0.4 mil when the director was being tracked optically and about 0.7 mil when the director was being tracked by radar. This reliability is certainly sufficient to assure dependable scores for radar tracking with the Mark 37 director where average tracking errors are about 10 to 20 mils. The method is inadequate, however, for the evaluation of Mark 37 optical tracking where average tracking errors may be less than 1 mil. Although this study demonstrates that it is impractical to try to obtain direct average er- ror scores for optical tracking, it should be pointed out that reliable average error scores can be predicted or estimated from time scores obtained by methods 1 or 2 above. Project N-lll evaluated this prediction procedure and compared predicted average error scores with the average of frame-by-frame measured er- rors. The predicted average errors had a probable error of only 0.1 mil for cases where a per cent time-on-target score had been ob- tained for a series of five courses and where the checksight scoring circle had been one of 3 mils diameter.10 Prediction is made on the basis of the formula: Per cent of time inside circle = 1—e~~^7 ... 4m2 where e = the base of the natural logarithm, r = scoring circle radius, and m = average miss or radial error. This formula assumes that tracking errors in traverse (or train) and elevation are normally distributed with the same standard deviation. That the assumption is justified for tracking with the gunsight Mark 15, at least, is demon- strated by the accuracy of prediction reported above. Average Error at Selected Bearing Points For the evaluation of tracking on crossing courses, Project SOS-6 tested the efficiency of getting an average tracking error score based on observed tracking errors at five preselected bearing points.19 (Obviously this method can be used only where the target is flying a pre- scribed course and the bearing points have some meaning in relation to the course being flown.) The correlation of scores obtained by this method with scores obtained by the timing method was +.84 for a series of azimuth courses by 25 different trackers. Reliability co- efficients, when two men scored the same course by the method, were approximately +.80. This is exceptionally good reliability and validity for a method as simple as this. At each of the five observation points, the checksight operators merely reported whether there was a zero error (if the radial error was less than 2 mils), a 2-mil error (if between 2 and 5 mils), a 5-mil error (if between 5 and 10 mils), or a 10-mil error (if over 10 mils). The five num- bers for each course were averaged to obtain the score for the course. Given any particular training task for track- ers, at least one of the above methods of check- sight scoring should be applicable. So that the training of personnel will no longer be based on unreliable, qualitative methods of rating trackers, the standardization of these or com- parable checksighting methods by the Services is of great practical importance. 3.2.5 a Phototube Scoring Device An instrument which would replace the checksight observer and obtain a time-on- target score mechanically is a phototube scoring device which was tested and improved for the Bureau of Ordnance by the Applied Psychology Panel.80 This device puts a lens, a scoring aper- ture, and a phototube in place of the checksight operator’s eye. A modulated light source of small area is mounted on the target. As the tracker tracks the target, the mechanism scores whenever the image of the light source falls within the scoring aperture and excites the phototube. Light intensity considerations require that the light source on the target be in a reflector and that the beam of light be directed at the EXPERIMENTS IN THE TRAINING OF TRACKERS 27 tracking device on the ground or ship. The source therefore has to be controlled by some- one in the plane who tracks, with no great precision (± 10 degrees), the place where the director or gun is located. If properly stabilized and operated, the device has remarkable range and sensitivity. Scoring should proceed satis- factorily out to ranges of 4,000 yards and may extend to 8,000 or 10,000 yards on clear days. Reliability of scoring with the device closely approximates the theoretical limit based on the change of effective light intensity from the searchlight with change of target range. This device was brought to its present stage of development only in the last days of the war and was not field tested. Nevertheless it de- serves very serious consideration as a training device for regular use on those sights and directors where the addition of the small weight and bulk of the recording unit can be tolerated. For disturbed line-of-sight directors, the photo- tube unit and lens must be coupled to a check- sight system and aimed at the target through the gunsight. For direct line-of-sight directors, the phototube unit can be attached to the track- ing head in any convenient spot. The operation of aligning the phototube aperture with the center of the tracker’s reticle is the same for either type of director system. Like checksight scoring, scoring with this phototube device has the advantage of provid- ing the tracker with a score immediately at the conclusion of each target run. The device may also be used to provide a signal during the run whenever the tracker is on the target. Such a signal can be an important aid in the training of trackers. 33 EXPERIMENTS IN THE TRAINING OF TRACKERS Although much has been said in the preced- ing paragraphs about ways of scoring tracking performance during training, no demonstration has yet been offered in this chapter of the specific value of such scoring in a training program. Experiments which provide this demonstration were performed by Project SOS-613’ 24 and Project N-114.12 3.3.x jjow Knowledge of Performance Aids Learning The effect of providing trackers with infor- mation concerning the accuracy of their track- ing during training was first studied in an experiment where the checksight observer used an electric buzzer which he sounded whenever the tracker was off the tracking point by more than 2 mils.13 Six men trained under these con- ditions were found to improve significantly more rapidly than six men trained without coaching. The conditions under which the “no coaching” group was trained resembled the conditions under which trackers frequently practice. The men took turns practicing on the Figure 10. Effect of buzzer coaching technique on tracking learning with the 40 mm gun. Scores are mean per cent of time off target on each of four tests. 40 mm gun using the computing sight M7 and no particular checks were made of their track- ing accuracy or of the adequacy of their drill- ing. The results of the experiment are presented 28 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROL in Figure 10. It can be seen in the figure that the group trained under the “no coaching” condition showed little or no improvement be- tween Test I and Test II, whereas the group trained with checksight and buzzer coaching showed rapid improvement. After Test II, the training conditions for the two groups were reversed. The group which before had “just practiced,” was now trained with the check- sight and buzzer. The men in the group showed prompt improvement and soon performed ap- proximately as well as the group first trained with the buzzer. It may be noted that the effect of the special training with the buzzer per- sisted for the first group for the duration of the experiment even though scoring was stopped for these men after Test II. These data thus show that the buzzer-training technique is effective in speeding up tracking learning. In a later experiment, the buzzer-training technique was compared with two other meth- ods occasionally used in training 40 mm gun comments before, during, and after each man’s practice on the gun. In the other method, the co-tracker guidance method, one of the gun pointers serves as instructor or coach as well as tracker, and calls “Off” whenever the other tracker makes errors. The results of this ex- periment are shown in Figure 11. It can be seen that the checksight and buzzer method of training, which provides the more immediate and more accurate tracking information to the gun pointer, leads to the best tracking, A similar experiment in radar tracking was not quite so successful as the two foregoing studies in showing the advantages of check- sight scoring.12 The experiment was run by Project N-114 at NTS, Fort Lauderdale, and concerned tracking with the radar Mark 4. The checksight used was that described in Section 3.2.3 for the gun director Mark 37. The scoring method was the method of average error out- lined in Section 3,2.4. Two groups of subjects were compared. One group was a control group. It consisted of four pointers and four trainers'3 who practiced but received no special training or coaching during approximately 8 hours of work on the Mark 4 (30 to 33 runs). Checksight scores were taken by the experimenters throughout the practice periods but were never reported to the men. The second group of subjects also consisted of four pointers and four trainers. These men were coached by checksight data for 12 runs and were then allowed to practice for 12 to 15 noncoached runs. Each session of noncoached runs, however, was begun with one coached run. During their practice, the men in the control or noncoached group showed only slight evi- dence of learning. There was little demon- strable increase in the tracking accuracy of the pointers and a greater, but still unreliable, in- crease in the tracking accuracy of the trainers. In the second group, the men trained as pointers became more accurate than the control group and improved in their tracking as long as coaching was continued. On removal of coaching, their scores became worse at first but Figure 11. Improvement in azimuth tracking performance in three groups of men trained by different methods on the 40 mm gun. pointers to track on target with the computing sight M7.22 One of these methods, called the verbal coaching method, involves coaching by the officer in charge. He makes suggestions and b In Section 3,3.1 the word trainer is used to mean a man who tracks in train. EXPERIMENTS IN THE TRAINING OF TRACKERS 29 by the end of the sequence of noncoached runs had improved considerably again. The trainers in this group, however, never became signifi- cantly better than the trainers in the uncoached group. For the Mark 4 pointers, the experimenters believed the training effects of the checksight scoring to be real. For the trainers, they attributed their failure to obtain more sig- nificant training effects to poor equipment maintenance, evidenced by generally large radar tracking errors observed during the tests. The study therefore points, as do many others, to the need for satisfactory equipment super- vision as a prerequisite to research on other variables. 3-3-2 Time Required in Learning to Track The data in Figures 10 and 11 provide some information on the time required to reach a probable final level of proficiency in tracking with the 40 mm gun. This is shown to be about of four men trained to track with the gun sight Mark 15 on the handle bar controlled gun director Mark 52.17 Their learning curves are shown in Figure 12. Each training period in- cluded some eight to ten courses. If an average curve for the group were drawn in the figure, it would show a plateau after about five train- ing periods. Actually, however, the data as shown in Fig- ure 12 are somewhat deceptive. Learning is not complete in five training periods. If perform- ance is considered as a function of range to target, it is found that the men were improving in their ability to track close-in targets right to the end of the training program. A break- down of average tracking performance for course sections when the targets were at dif- ferent ranges is given in Figure 13. These data Figure 13. Variation in tracking accuracy with target range. Data averaged for four men track- ing aerial targets from a stationary platform using the gunsight Mark 15, suggest that about two weeks of training, ten courses per day, would be required to train men to the point where they stay on target (within 1.5 mils) more than 90 per cent of the time for targets at short range. This refers to tracking from a stable platform. Tracking from a rolling deck is more difficult, and data paralleling the above indicate that about 200 courses of aerial Figure 12. Individual learning curves for four trackers on the gunsight Mark 15. 50 or 60 courses if each course is of about 1 minute’s duration. Comparable data were obtained for a group 30 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROL target tracking drill are needed to bring track- ers to their learning plateau for operation on a rolling deck. 34 FACTORS TO CONSIDER IN THE TRAINING OF TRACKERS This discussion of the training problems of trackers would not be complete without raising some questions relative to a number of training principles which may well determine the pat- tern for and the success of later training pro- grams. Selection of Personnel The selection of personnel to be trained as trackers, as for any type of job, can be expected to reduce training time and improve the final level of operator performance. For some tasks, it is efficient to use perform- ance on early trials with the equipment itself as a means of selection or screening. This is probably true of tracking. One indication of the success of this selection procedure for tracking is found in the report of one of the Foxboro Laboratory studies.7 Average per- formance scores over successive ten-trial train- ing periods were obtained for each subject. Correlations between scores on each of the earlier sets of trials with scores for the sixth set of ten trials were determined. These corre- lations are shown in Table 1, The correlation of +.68 between the scores on the first block of trials (I) and those on the sixth (VI) indi- cates that in the training of these subjects useful judgments could have been made from training trials are found in test data which showed that the Tufts tracking trainer was a good selection device for M7 director track- ers.9 Thus, for optical tracking at least, it appears that a working method of selecting operators would be to use checksights to administer pre- liminary tracking tests to those men who are available for assignment as trackers. 3‘4'2 Training Goals When men have been selected and are under- going training as trackers, it is important that certain goals in tracking performance be set and that the men check daily on how they are progressing toward those goals. Inasmuch as tracking accuracy varies with so many condi- tions (target range and angular rate of mo- tion, roll and pitch of the deck, the director or gun in use, the kind of tracking control, etc.) it is obviously necessary that these training goals be made specific for each tracking device and for the conditions under which it is used. One attempt at setting such tracking standards based on assembled tracking data was made in a report by Project N-lll.17 The data, ap- plicable to the gunsight Mark 15, are repro- duced in Table 2. Reprinted in several operat- Table 2. Acceptable tracking proficiency after 2 weeks of tracking training with the gunsight Mark 15. For tracking at ranges between 4,500 and 1,500 and under 40° elevation Per cent of time with tracking error less than 1.5 mils Approximate average error equivalent (mils) Tracking aerial targets from a stable deck 90-95 0.9-0.8 Tracking surface targets from a rolling and pitching deck (5° to 10° motion) 70-80 1.2-1.0 Tracking aerial targets from a rolling and pitching deck (5° to 10° motion) 50-60 1.6-1.4 Table 1. Rank order correlations between aver- age scores for groups of trials during pointer matching training (20 subjects). I* and VI II and VI III and VI IV and VI V and VI + .68 + .77 + .85 + .87 + .92 * All Roman numerals refer to blocks of ten trials except the number I. In this block only seven trials are included, three trials were considered practice. since the first mg pamphlets, these standards serve as a guide to those officers who are responsible for train- ing men and as a performance goal for the men in training. early trials as to which men were likely to be superior at the close of training. Similar im- plications for the predictive value of early FACTORS TO CONSIDER IN THE TRAINING OF TRACKERS 31 3 4 { How Well Is Tracking Skill Retained? Data on the retention of tracking skills are important in determining the need for re- fresher drills and tracking exercises. Some research reports contribute information on this point. Two groups of men trained in azimuth track- ing on the 40 mm gun under conditions of checksight and buzzer training were tested after an interval of no practice to measure the skill they retained.25 These data are summar- ized in Figure 14. It can be seen that the par- curacy after from 16 to 29 days of no prac- tice.7 Pending the accumulation of additional data, the conclusion seems indicated that once men are adequately trained in tracking they need refresher drills no more often than once a month. 3.4.4 jg bracking Skill a General Skill? The Services have occasionally expressed in- terest in the design and construction of a gen- FlGURE 14. Curves showing (A) the improvement in tracking performance and (B) the retention of track- ing skill in groups of 40 mm gun pointers. tially trained trackers, the men of Group 1, showed no decrement but actually showed some improvement when part of the group was tested after two, three, and four weeks of no practice. In the case of the highly trained men, Group 2, six of whom were tested after one week and six after 2 weeks with no practice, some but little loss in performance was ob- served. A Foxboro Laboratory experiment presents similar data indicating no loss but actually some improvement in tracking ac- eralized tracking trainer, an instrument which could be used to train trackers for all or many types of tracking tasks. This suggestion raises a more fundamental question, however, as to whether tracking skills are specific to particular devices or whether they are general. If they are general, then men may be transferred from one tracking job to another with no doubt as to their ability to do the second job as well as the first. A general trainer would be feasible. But if tracking skills are specific to particular 32 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROL Figure lb. Tufts tracking trainer with housing removed, front view. TRAINING AIDS FOR OPTICAL TRACKING 33 instruments, men must be trained for each tracking job in a particular way and on a reasonably specific trainer. An approach to this problem was made in an experiment by Project SOS-6.15 The project recruited six small groups of trackers who had learned on-target tracking. The groups com- prised men who had been trained, respectively, on the director M7, director M5, Tufts tracking trainer, heightfinder, radar SCR-268, and the 40 mm gun. The groups were all given a short test for tracking skill on the Tufts (director M7) tracking trainer. Except for the groups trained on the director M7 and the tracking trainer itself, none of the groups were superior to a group of untrained men tested on the same instrument. These data argue against the no- tion of generality and suggest that training on one tracking task does not necessarily pro- vide skill in another tracking task. However, other data show that as long as a man continues to work at the same instrument there is a high degree of carryover of skill from tracking one course to tracking another with similar com- ponents7 and from tracking with one form of control to tracking with another.28 Either of two explanations is possible for these data. Either the degree of transfer of tracking skill from one tracking device to an- other depends specifically on the resemblance between the two tasks and the number of identical coordinations they involve, or the study by Project SOS-6 concealed a general skill factor by failing to follow the men long enough to see how they would have compared in ability once they had become completely familiar with the new job. It is important that the Services know which of these alternatives is correct; the answer governs a number of important decisions with respect to training and the assignment of per- sonnel. 35 TRAINING AIDS FOR OPTICAL TRACKING The role of trainers in a program of training men to track aerial targets can be a significant one because of the flying time, fuel, and general operational costs which are saved and the in- dependence of weather conditions which is gained if the training of trackers can be car- ried out without having to have a target plane in the sky. Several Applied Psychology Panel projects assisted in the trainer phase of the optical- tracking jobs, in the experimental evaluation of trainers, and in the preparation of instruc- Figure 16. Tufts tracking trainer, rear view. tion guides and lesson plans to be used with standard trainers. 3-51 The Development of Optical Trainers Three on-target tracking trainers were de- veloped by Panel projects. These were a trainer which simulated the task of optical tracking found on the stereoscopic heightfinder,2 a trainer which simulated the two-man tracking task in the gun director Mark 37,20 and a trainer simulating the two-man tracking task 34 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROL on the director M7.6 All these trainers were similar in principle in that they scored by re- cording differentially the cam-controlled mo- tion of the target and the tracking input of the operator. A discussion of the M7 trainer only is included here because this trainer was the most elaborate of the three devices and the one which was evaluated most systematically. Built by Project SOS-6, under a Tufts College contract, it became known as the Tufts track- ing trainer. In size and in general appearance (with the exception of the added target units) the Tufts tracking trainer resembles the director M7 very closely. The critical factors of tracking- telescope position and handwheel position, size, torque, etc., match those of the director M7. The azimuth and the elevation tracker use direct handwheel tracking to keep the reticle crossline in the telescope on the tracking point of the target. The trainer is shown in Figures 15 and 16. When the azimuth handwheel is turned, the whole trainer turns on the pedestal, as would the director itself. When the elevation hand- wheel is turned, the telescope tubes in the target unit move in the same way that the tracking telescopes elevate or depress on the director. The target is presented on a Kodachrome slide. Courses are regulated by large cams installed in the body of the trainer. A recording unit produces a graphic record of tracking accuracy. A “time-off-target score” or a “per cent of time- off-target score” can also be obtained for each course. ! ’2 The Experimental Evaluation of Trainers The Tufts trainer was tested9 for validity as a training device and, in a preliminary way, for validity as a selection device (Section 3.4.1). Thirty-two enlisted men were used in the training experiment. After a preliminary tracking test on the Tufts trainer, the men were divided equally into two groups so that the groups were similar in tracking skill. One of these groups was trained in tracking on the trainer while the other was trained on the director M7 itself. The men being trained on the trainer learned after each trial how well they had tracked. They were told what amount of time they had been off target by two mils or more and were given a graphic record of their tracking performance. The men trained on the director tracked airplane targets and were put through standard tracking practice sessions (without checksight scoring) for an amount of time equivalent to that devoted by the other men on the trainer. After approxi- mately 2V2 hours of actual tracking practice per man, the two groups were tested for their skill in tracking on the director M7 itself. The director which was used was equipped with a recording camera, and tracking photographs were taken, one per second, during the test. In order to permit a further comparison of the groups, all the men were given a post-training test on the trainer. The results of these tests are presented in Table 3. The data show that the group trained on the trainer performed somewhat better on the director in the post-training test than did the group trained on the director. The differ- ence between the groups in director-tracking skill approached statistical significance (it might have occurred by chance about one time out of ten). The records for the final test on the trainer confirmed this group difference. The observed superiority of the trainer group in the post-training tests is probably to be attributed to the greater information concern- ing errors received by this group and to the basic similarity of the trainer to the director in those characteristics essential to the track- ing operations. Although the Tufts tracking trainer was shown by these results to accomplish a real training job, it was not adopted for regular Army use because it was not ready until too late in World War II. The trainer will probably be of continuing interest to the Services, how- ever, because it solves a number of mechanical problems not dealt with successfully in other trainers. Among these is the problem of pre- senting the target in such a way that the men are required to walk around with the trainer while tracking in azimuth, just as they do when tracking a real target with actual director equipment. TRAINING AIDS FOR OPTICAL TRACKING 35 One aspect of trainer evaluation relates to the difficulty of the trainer task in terms of learning time. Learning curves for the Tufts trainer, obtained for the trainer group in the that reached by the same men in tracking similar courses on the director itself. Learning time on the trainer was similar to that for tracking with the gunsight Mark 15 from a Table 3. Results of training test using the Tufts tracking trainer. Scores are given as average errors in mils. Mean score for Mean score for •n Probability of the difference occur- ring by chance Tests men trained on the M7 men trained on the trainer mils Pre-training test on trainer (basis on which groups were matched) 2.86 2.85 0.01 50/100 (t = 0.02) Post-training test on trainer (same test as pre-test) Difference: pre-training test vs final test post-training Post-training test on director. Azimuth scores 1.48 1.38 (t = 4.81) 1.20 1.65 (t — 4.44) 0.28 9/100 (f = 1.40) Trial 1 1.12 0.79 0.33 Trial 3 1.08 0.95 0.13 12/100 (f = 1.19) Total 1 + 3 Elevation scores 1.10 0.86 0.24 Trial 1 0.86 0.71 0.15 Trial 3 1.22 0.91 0.31 Total 1 + 3 1.03 0.81 0.22 9/100 (t = 1.36) stable platform (Section 3.3.2). It was similar also to the times indicated in Figure 18, based on two studies of pointer-matching training performed at the Foxboro Laboratory.7*8 The curves in Figure 18 indicate that 35 to 40 trials are required to reach a performance plateau at this job. This turns out to be about 150 minutes of actual practice time. A second optical tracking trainer which was studied experimentally was the gunnery trainer Mark 5 designed for use on the gunsight Mark 14. When this instrument was in pilot- model form, the Bureau of Ordnance requested Project N-lll to carry out an evaluation study. Early tests made by the project indicated that the scoring device incorporated in the trainer was too unreliable to provide meaningful train- ing scores, so the investigation turned to a study of ways of improving the scoring system. The source of unreliability was found to be in the phototube system and was compensated for by increasing the intensity of the target image on the phototube and increasing the elfective sensitivity of the “target image in phototube aperture” system.39 The changes required by these improvements were made in subsequent production models of the trainer. Tests of a Figure 17. Average tracking error for azimuth practice trials of 16 men trained on Tufts track- ing trainer. test just described, are shown in Figure 17. They reached a plateau in about 2 hours of actual tracking practice spaced over a 2-week period. The average tracking error score reached on the trainer was slightly lower than 36 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROL number of the latter have shown that scoring in units coming “off the line” is dependable. Because of the delays which a thorough valida- tion study would have introduced in the produc- tion program, the gunnery trainer Mark 5 was guides, lesson plans, and drill outlines for work with these devices.21’31’35 In the preparation of the drills an effort was made to (1) maximize time available for drill and minimize time spent in lectures and dis- cussion; (2) minimize “dead” time when men were not profitably employed; (3) provide for standardized and adequate coaching; (4) util- ize charts, diagrams, etc., as illustrative teach- ing devices; and (5) introduce drills with mov- ing targets as early as possible in the training programs. A sample of one of the drills is shown in Figure 19. The sample is for the panoramic trainer, Mark 12. Other trainers for which lesson plans and drills were prepared included the portable aiming teacher, the multiple for- ward area sight trainer, the Mark I trainer (Polaroid, machine gun trainer), the Mark III trainer, Model 1 (for forward area sight train- ing), and the Mark VI trainer (fixed gunnery deflection trainer). 36 RADAR TRACKING TRAINERS Work with radar tracking trainers was undertaken by Projects SC-70, N-lll and N-114. SC-70 and N-114 contributed to the de- velopment and testing of trainers which simu- lated the pip-matching tracking task of the radar Mark 4. Projects N-lll and N-114 de- signed devices to simulate tracking with a radar which provides a T and E dot. 3-61 'pjie Development and Evaluation of Two Pip-Matching Radar Trainers The Foxboro trainer (BC-968-A) is a device for training men to track by pip matching. It is an electronic-mechanical unit which feeds synthetic target echoes into an indicator unit for SCR-268 radar equipment or other equip- ment types using pip matching. All scope con- trols are reproduced, and as the operator tracks on a cam-controlled course a Veeder counter score and a tape record of his tracking perform- ance are taken. In one evaluation test for this trainer,4 25 MEN TRACKERS EXPERIMENT I«-*(N = IO) REFERENCE 7 EXPERIMENT!! •—•(N = 6) REFERENCE 8 Figure 18. Learning pointer matching, a plot of tracking error in degrees handwheel rotation by trials. ordered into production on its face validity alone. Now that a recording camera is avail- able for use on the gunsight Mark 14 (Section 3.2.2 above), the validation study of the gun- nery trainer Mark 5 should be undertaken. 3.5.3 Preparation of Instruction Guides and Lesson Plans for Use in Trainer Programs There were a number of optical-tracking trainers, among them several lead-tracking trainers, which were adopted by the Services early in the war and which were put into regular use at training stations throughout the country. In order to make training with these devices as effective as possible, Project N-105 assumed the job of preparing instruction TRAINING AIDS FOR OPTICAL TRACKING 37 I PURPOSE: To give practice in maintaining skill in use of forward area sight on combat targets. II CONDUCTED BY: Enlisted instructor. Ill PREPARATION Materials needed: Three Mark IV trainers per crew. A. Lecture (2 min). Know contents A, part IV. B. Drill: See that trainer is in running order. IV PRESENTATION A. Lecture (2 min) : Nature of this drill. 1. The purpose of this drill is to give the men practice in keeping up their skill in the use of the forward area sight on combat targets. Tell men Mark IV trainer is commonly known as panoramic. 2. The men will fire a 20 mm gun at moving targets. a. The gunner will be strapped to the shoulder bars. b. He will put his head into the eyepiece and using both eyes he will see (1) motion pictures of planes going through combat maneuvers, (2) a forward area sight which can be moved, (3) a red flash when a hit is scored. c. The gunner will have to track the target and to sight the target cor- rectly. d. A trunnion operator will raise and lower the column for the gunner. 3. The gunner will fire by squeezing trigger on left handgrip as follows: Point out trigger. a. Open fire as soon as plane appears. b. Fire in bursts until hits are made. c. Fire continuously thereafter until plane flies off screen. 4. The speed of the plane will be between 200 and 300 knots. 5. The number of hits will be recorded automatically. Number of rounds fired is also recorded but don’t tell crew. 6. The film will stop automatically. B. Drill 1. Assign the men as follows: a. First man is gunner. b. Second man is trunnion operator. c. When firing is finished men reverse positions. d. When first pair finishes, a second pair takes positions. 2. Coaching a. Stance (1) See that feet are placed well forward. (2) See that gunner is balanced on both feet. (3) See that gunner walks with gun. b. Firing (1) Remind gunner that lead is constantly changing except for zero approach angles. (2) Don’t allow long periods to go by without the gunner firing. c. See that trunnion operator works column quickly and smoothly to keep gunner’s knees slightly bent. d. Scoring. (1) Tell gunner and rest of crew number of hits made but not number of rounds fired. (2) Record number of hits and number of rounds fired. 3. Before dismissing men look over the scores and comment on the men’s per- formances and difficulties. Figure 19. Mark IV (3A-11) trainer, one drill, forward area sight, combat targets. 38 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROL enlisted men with no previous experience in radar tracking were given 9 minutes and 36 seconds of practice per day (three runs of 3 minutes 12 seconds each) for 12 successive days. One week following their last training trial, they were tested to see how much skill they had retained. Curves showing the improvement that took place in the average performance of the men during the training days are shown in Figure 20. It can be seen that by the twelfth day, after men for Mark 4 radar tracking. The experi- ment was run at the Naval Training School, Fort Lauderdale, Florida, by Project N-114. Concurrently, a test was made of the validity of an optical-mechanical pip-matching trainer designed jointly by Projects SC-70 and N-114.18 In this device, the scope presentation is accom- plished by passing light through two pip- pattern slits and green filters and having the light fall on a ground glass surface which simulates the scope face. The remainder of the apparatus consists of (1) a cam device to unbalance the pips (i.e., to make one higher or lower than the other) in a manner resem- bling pip variation when a target is flying crossing courses; (2) handwheels, whereby the operator can offset the tendency of the course cam to unbalance the pips and thus keep them matched in height; (3) a jiggle mechanism, which causes movement of the light source to produce a complex bobbing of the pips, there- by increasing the difficulty of the task and in- troducing a factor found in tracking with Mark 4 radar; and (4) a scoring device, which provides both a graphic record and an inte- grated number score indicative of the profi- ciency of tracking. In an experiment to evaluate the two train- ers, 36 men previously untrained in radar tracking were used. At the outset, they were given a pretest in tracking aerial targets (pointing and training) with the radar Mark 4 in the gun director Mark 37, On the basis of their pretest scores they were divided into three matched groups. One group of 12 men was trained for 12 days on the Foxboro trainer (three trials daily, each trial of 3 minutes duration). The second group of 12 was trained for 12 days on the optical-mechanical pip- matching trainer (three trials daily, each trial of 2 minutes 15 seconds duration). The third group was given no radar tracking practice during the 12-day period. For the groups trained on the two trainers, daily progress was recorded on a posted score-sheet, so that sub- jects could compare their daily performance with their past records and with the perform- ance of other subjects. This contributed toward keeping motivation high in the two experi- mental groups. Figure 20. Learning curve showing development of proficiency with daily practice on Foxboro trainer (N = 25). approximately 115 minutes of practice, the curve had begun to level off into a plateau. Variability of performance from trial to trial as measured by the standard deviation on each day decreased continuously, as shown by the dotted curve. There is little change in either accuracy or variability after one week without practice. In a later experiment,14 one of these Foxboro trainers was modified so that the tracking con- trols resembled the tracking handwheels in the gun director Mark 37 and so that the scope face was in the same position relative to the tracker as in the Mark 37. The instrument was then evaluated for its efficiency in training RADAR TRACKING TRAINERS 39 Analysis of the learning curves for the men trained on the Foxboro trainer con- firmed the conclusion drawn from Figure 18 above, namely, that performance on the trainer reaches a plateau in about 12 days of practice. For the optical-mechanical pip- matching trainer, however, almost all improve- ment in performance occurred during the first 3 days of practice, with slight improvement continuing until the eighth day. niques which should be employed with the device.3 Table 5. Radar tracking training with two pip- matching trainers. Statistical comparison of ex- perimental and control groups when pointing and training during post-tests are considered sepa- rately. Pre-test Post-test Experimental post- test vs control post- test Foxboro trainer: M t V Pointers 18.4 15.5 7.2 3.6 .004 Trainers 14.7 10.9 3.4 2.1 .06 Control: Pointers 18.9 22.7 Trainers 14.5 14.2 Optical-mechanical trainer: Pointers 20.5 17.3 5.4 2.5 .03 Trainers 12.6 9.8 4.4 3.4 .006 Table 4. Radar tracking training with two pip- matching trainers. Comparison of combined point- ing and training average error scores for experi- mental and control groups in tests for tracking- on the director itself. Group trained on Control Group trained on optical- mechanical Foxboro group trainer (N=12) (JV= 12) (N= 12) Pre-test group mean 16.6 mils 16.7 mils 16.6 mils Post-test group mean 13.2 mils 18.4 mils 13.4 mils Experimental post Mam = 5.2 mils M = 5.0 mils test vs control t = 3.8 t — 3.9 post-test V — .003 p = .002 362 The Development of Two T and E Type Radar Trainers Two other radar training devices which the projects worked on were devices which simu- lated T and E type radar tracking. The first was a mock-up of the dot tracking job on the radar Mark 12. This tracking pres- entation was incorporated in the optical track- ing trainer for the gun director Mark 37 (Sec- tion 3.5.1 above), built by Project N-114.43 In an experiment to determine the validity of this instrument as a tracking trainer for the radar Mark 12, practice was found to result in significant improvement in tracking on the trainer, but men so trained showed no superi- ority on the Mark 12 itself over a control group which had received no training of any sort. The device, therefore, had no validity as a trainer for the radar Mark 12. These results should serve to emphasize the caution with which one must regard proposed trainers. Practice on a device which has a good amount of face validity and which results in a large amount of improvement in performance on the device itself does not necessarily transfer to the real job on combat equipment. Accepting the view that a trainer does not merit the name until it has been proved valid At the end of the training period, the three groups were again tested (pointing and train- ing) in the gun director for their ability to track aerial targets with the Mark 4. In these post-training tests, the two groups trained on the trainers tracked with greater accuracy than did the control group which had had no train- ing. When pointing and training scores were combined, the performance differences in favor of the trainer groups were statistically sig- nificant for both trainers. These data are pre- sented in Tables 4 and 5. The experimental data show, therefore, that both the Foxboro instrument and the optical- mechanical pip-matching device are valid train- ers for the operations of tracking with the radar Mark 4. The Foxboro trainer, in pro- duction early in World War II, was already in use at many training stations. To aid the Serv- ices in obtaining effective training with this trainer, Project SC-70 prepared a special report summarizing the methods and training tech- 40 TRAINING TRACKERS FOR ANTIAIRCRAFT FIRE CONTROL for training, Project N-lll called the instru- ment which it developed a radar tracking demonstrator,49 The instrument is a device which generates a dot type of radar signal and simulates the presentation on the pedestal-type, radar-equipped gun directors, the Marks 57, 60, and 63. The synthetic radar signal, which re- produces the appearance of an actual signal with considerable fidelity, is fed to a T and E scope mounted on the tracking head of a stand- ard pedestal director. In tracking, the man moves around the director and performs all the operations required in following a real target with the real radar system. The signal generating unit, with cams to control courses and a scoring system to register performance, may be mounted on the director in place of the gunsight normally positioned on the tracking head, or it may be located away from the director and used to feed signals to the regular T and E scope in the standard gunsight on the director. This demonstrator was completed just at the close of the war and is now undergoing further test and development at the Naval Research Laboratory. 3 7 TRAINING AIDS AND TRAINING MATERIALS In addition to preparing trainers and study- ing training methods, several of the projects under the Applied Psychology Panel gave further assistance in the general program of training trackers by developing training aids and training materials of various sorts. Their work included assistance in planning courses of instruction for gunners, trackers, and di- rector operators, the preparation of course out- lines, lesson plans, and drill outlines, and the writing of pamphlets on operating technique and theory. Illustrative of this work are the following. Lesson plans for courses of instruction in 20 mm gunnery,1 40 mm gunnery,3-32 and in the operation of the gun directors Mark 37,10 Mark 57,44 and Mark 63.42 Drill manuals for the use of gunnery train- 0X*S Operational pamphlets, discussing tracking in relation to the general operation of pedestal-type Navy gun directors.36’40’41-46 Achievement test items for courses of in- struction in director operation.37-43 Material on fire control nomenclature34 and monographs on training theory and meth- ods47 and on the theory of lead computing systems.48 These instructional aids were useful in im- proving the training of trackers, in increasing their understanding of the work they per- formed, and in developing standardized meth- ods of operation. Chapter 4 TRAINING PERSONNEL IN RANGE DETERMINATION William E. Kappauf, Jr* SUMMARY The range to a target may be measured stereoscopically or by radar, or it may be estimated either with or without the use of a stadiametric aid. Projects of the Applied Psychology Panel suggested a few improvements in stereoscopic trainers, but their major contribution was in the preparation of training manuals for in- structing Army personnel in heightfinder op- eration. The same material was in large part adapted for the training of Navy personnel in rangefinder operation. To improve training further, a number of training aids and teaching aids were developed. The rate of improvement during training in heightfinder and rangefinder operation was studied in a variety of situations. About 2,700 trials were found necessary on the M2 trainer and about 3,200 on the Eastman trainer before men reached a plateau. Men who had this amount of training on the trainer reached a plateau in their ability to range on aerial tar- gets in about 500 trials. Men completing heightfinder or rangefinder training showed a mean deviation of about 2.5 units of error [UOE] in aerial target height readings and a mean deviation of from four to five UOE in range readings on aerial targets. Training on the firing line with the reticle of the gunsight Mark 14 was found to be su- perior to training on the mirror range estima- tion trainer device 5C-4 for learning stadia- metric ranging. Unaided estimation of the opening range for 20 mm fire, 1,700 yards, was found to be superior to stadiametric ranging with the gun- sight Mark 14. Practice in aerial target range estimation for ranges up to 8,000 yards increases the per cent of estimates within 15 per cent of true range from about 25 per cent to 40 or 45 per cent. Relatively frequent refresher drills are necessary, however, to keep men to this level of performance. The accuracy of estimating the range of aerial targets is approximately the same as the accuracy of estimating range of ground targets. 41 METHODS OF RANGE DETERMINATION In the operation of antiaircraft batteries, range to the target is determined by any one of four methods: radar, rangefinder, some type of visual stadiametric device, or estimation. Precision of the methods of determining range decreases in the order given. The fact that the last two methods are less accurate than the first two does not mean that they are any the less important. A number of fire-control sys- tems, directors, or sights, used at short range or delivering only approximate solutions to the firing problem, regularly depend on stadia- metric or estimated ranges in standard operat- ing procedure. The Applied Psychology Panel was called upon to assist in the training of operators for the most precise and highly developed of the Army and Navy director systems as well as to assist in the training of gunners and gunsight operators who used less refined range data. Panel projects contributed training aids, train- ers, lesson plans, course outlines, and examina- tions. They also conducted experiments to evaluate the efficiency of particular training methods and to determine the standards of proficiency to be expected of range men in a battery or at a gun. The selection of rangefinder personnel is described in Volume 1, Chapter 8, the Summary Technical Report of the Applied Psychology Panel, and the general procedures of range- finding in Chapter 22 of this volume. Radar training and radar operation are described in Chapters 2 and 21 respectively. The selection of radar operators is covered in Chapter 7 of Volume 1. a This chapter is based on work by NDRC Projects N-105, N-lll, N-114, and the Height Finder Project and on earlier work by NDRC Division 7. 41 42 TRAINING PERSONNEL IN RANGE DETERMINATION 4.2 XHE TRAINING OF STEREOSCOPIC RANGEFINDER AND HEIGHTFINDER OPERATORS 4.2.i The Development of Stereoscopic Trainers The need for a stereoscopic training instru- ment was recognized a good many years ago. Under Army and Navy contracts, the stereo- scopic trainer M2 (or Mark 2 in Navy termi- nology) was developed. This is a relatively simple trainer and does little more than present the target and reticle patterns in a viewer re- sembling a stereoscope. Provision is made for changing the target silhouette. It is also pos- sible to introduce changes of target range and of its azimuth and elevation positions, so that the operator experiences some of the diffi- culties of ranging on an aerial target. A more versatile and realistic trainer was developed by the Princeton University group.1 Known originally as the Eastman trainer, this device uses Kodachrome target slides, intro- duces tracking errors and range changes by cams, and makes a record of the observer’s errors in ranging. The Princeton University group tested these two instruments to determine their validity as trainers.1 A class of 36 students at the Height Finder School, Fort Monroe, was divided into three sections. One section was trained by the usual methods on the heightfinder, one section was trained on the M2 trainer (modified to include a motor range device), and the third was trained on the Eastman trainer. Both trainers proved to be valid. When the men of all sections were tested on the heightfinder at the end of 4 weeks, 7 weeks, and 10 weeks of training, their performance as measured by a variability or precision score was equivalent. The men who had had 7 weeks of work out of 10 on the trainers made readings which were no more variable than the readings of the men who had practiced through the entire course on the heightfinders. The Height Finder Project and Project N-114 of the Applied Psychology Panel entered this field after the above developments were complete. Its trainer contributions to the stereo- scopic training program were limited to the following. (1) A simplified motor drive was designed to control an automatic range input on the stereoscopic trainer M2.4 (2) A method was developed for controlling both range and tracking errors by motor drive on the Navy trainer Mark 2.16 (3) Some preliminary work was done on stereoscopic spotting trainers. Many such trainers have been built for the purpose of teaching rangefinder operators how to spot burst errors in range, but few, if any, have ever been validated. At the request of the Navy Bureau of Ordnance, Applied Psychology Panel projects submitted a design for a better spotting trainer3 and submitted criticism of a proposed polaroid spotting trainer.28 What seemed to be needed most in handling the spotting problem, however, was not a trainer for the present spotting job but a way of in- troducing range spots in UOE instead of in yards. Until this instrument development is achieved, stereoscopic spotting will always be inaccurate because of its dependence on an estimate of the absolute range of the target. 4.2.2 The preparation of Training Manuals A manual on the theory, operation, and maintenance of heightfinders was prepared for use at the Height Finder School, Camp Davis, North Carolina.6-7’9 The manual was first dis- tributed in mimeographed form. Later the prin- cipal chapters of the manual appeared in a new Army pamphlet on the heightfinder,38 while other parts were incorporated in field manuals and ordnance manuals.32- 33>34 The manual was written so that the new heightfinder operating procedures developed by Division 7 and the Height Finder Project would be available to heightfinder operators. Important among the procedures established in the manual were the following suggestions. 1. That the interpupillary adjustment be made with the aid of an interpupillary tem- plate. 2. That the height-of-image adjustment be made on a target at a range greater than 2,000 yards and with the measuring scale set to more than 2,000 yards. 3. That the internal adjustment, when made TRAINING RANGEFINDER AND HEIGHTFINDER OPERATORS 43 on the internal target, be made at 650 mils elevation and with the height range lever in height. 4. That the internal adjustment, when made on a star or on the moon, be made with the height range lever in height. 5. That the calibration correction be com- puted on the basis of combined data for the fixed target, aerial target, and celestial target readings, omitting data on any targets at ranges shorter than 2,000 yards. 6. That the wedge check be carried out with the elevation tracking telescope on a carefully surveyed “level point.” 7. That the wedge check be carried out at height-900 yards instead of height-550 yards. This recommendation simplifies the wedge check procedure and is in line with the new manufacturer’s inspection procedure. 8. That the wedge adjustment be considered satisfactory when the difference between range-infinity readings and height-infinity readings does not exceed two UOE and when the difference between range-infinity readings and height-900 readings does not exceed one UOE. 9. That regular measurements be made of the amount of backlash between the main bear- ing race and the bevel pinion and that ordnance repairs be made to reduce it. 10. That vertical slot-shaped end-window stops be used for observing at 24 power except when the illumination is too low to permit it. When Project N-114 was set up at NTS, Fort Lauderdale, members of the project assisted the school staff in adapting parts of the Army pamphlet to Navy use. Most operat- ing and training procedures, except those spe- cific to the Army heightfinder, were carried over. Later, members of the project assisted in writing a revision of the ordnance pamphlet on the care, adjustment, and operation of stereoscopic rangefinders.23 4-2-3 The Development of Training Aids and Teaching Aids A considerable number of training aids and teaching aids were developed by the Applied Psychology Panel for use in the training of stereoscopic rangefinder and heightfinder op- erators. Some of these required background research, but most of them were prepared and recommended for school adoption on the grounds that they represented good teaching technique in the eyes of the project staff, A list of these training aids follows. 1. Simplified record forms for observers to use in scoring their performance, in computing their calibration correction, in checking the zero mils adjustment of the heightfinder, and in noting what maintenance work is done on their instruments.7 2. A short pamphlet of arithmetic instruc- tion, later included in the heightfinder manual.7 3. A series of classroom wall charts on the construction of the heightfinder and range- finder.14 4. A detailed course outline for a 12 weeks’ course of instruction in the operation and maintenance of the heightfinder.9 5. A series of worksheets to parallel the chapters in the new heightfinder manual.14 6. Simplified models of the optics of the heightfinder5 and rangefinder.23 7. An interpupillometer for the accurate measurement of interpupillary distance.15 8. Interpupillary templates to be used in making interpupillary distance settings on a rangefinder or heightfinder.14 9. A range correction computer for deter- mining a parallax correction for the range to the target from each of a series of rangefinders used in a row at a training station.12 10. A rangefinder slide rule, which was a simplified rule for computing rangefinder errors in units of error.13 11. A set of achievement test questions and performance test items for operators of Navy fire-control equipment.29 12. A set of lesson plans for an officers’ course on the selection and training of range- finder operators.17 13. A set of instructions and suggestions to heightfinder instructors.9 42 4 Special Suggestions for Training The specific importance of a good calibration correction procedure for aerial target read- 44 TRAINING PERSONNEL IN RANGE DETERMINATION on the graphs which follow are corrected by a factor of 2 wherever variability scores are used. Curves showing the decrease in the varia- bility of student readings with practice on the heightfinder are shown in Figures 1 and 2. ings,10 and of a precise height-of-image adjust- ment for taking infinity readings on stars,8 was shown experimentally. These two matters are discussed in detail in connection with the development of better heightfinder operating procedures in Chapter 22. Their implications for a training program are these: students need to be thoroughly trained in the techniques of calibration which they may have to use when they are in the field. They should be drilled in infinity-target readings and ground-target readings until they can get a calibration by these methods which is just as satisfactory as a correction based on aerial target readings. They should also be drilled in height-of-image adjustment technique until they can make this adjustment with a precision (spread less than 1 mil of apparent field) which does not destroy the accuracy of infinity readings on stars. A further test result of importance to train- ing was the demonstration that Mark 10 radar ranges on targets beyond 6,000 yards were satisfactory reference ranges for use in scoring student operators or in calibrating rangefind- ers.21 The method requires, of course, an accu- rately calibrated radar but it has the great advantage that it lends itself to use aboard ship and in the field where calibration data are usually hard to obtain. 4 2 5 Learning Rates There are five studies from which learning data may be drawn in an analysis of the time required to train stereoscopic operators.1’2> 16) 20, 27 Two measures of student performance were used in these studies: the UOE score and the variability score. Both are measures of the variability of observer readings. The UOE score is the mean deviation of the readings expressed in units of error; a unit of error is a disparity of 12 seconds of arc at the observer’s eye between the right and left eye field of view. The variability score is a measure of the spread of the readings. It is based on the median errors of sets of three consecutive readings. The varia- bility score turns out to be between 2 and 2.5 times as large as the UOE score, so the scales I I i I FORT MONROE DATA CLASS 8, N= 28 CLASS 9, N = 28 CLASSIC, N= 2 7 CLASS 12, N= I 2 Figure 1. Learning curves showing decrease in variability of fixed target readings on the height- finder. (Classes had no previous training on the trainer.)1 None of the classes represented in these data had any previous or concurrent training on stereoscopic trainers. Figure 1 indicates that 3 weeks of training in ranging on fixed targets is sufficient to bring the students to their best CLASS 12 N = 12 Figure 2. Learning curve showing variability in aerial target altitude readings on the heightfinder. (The class had no previous training on the trainer.) x> 27 level of performance. Though the data in Fig- ure 2 are based on the records of a very few students, they show that the typical UOE score for height measurements by graduating stu- dent observers is between 2 and 2.5 UOE. TRAINING RANGEFINDER AND HEIGHTFINDER OPERATORS 45 Since most of the heightfinder and range- finder operators trained in schools during the war received their initial stereoscopic training on synthetic devices, Figures 3 and 4 are of interest. In the heightfinder training program, the first few classes trained on trainers were required to make a great many trainer read- ings on both fixed and moving targets. Learn- ing records for these classes were analyzed to determine how many practice readings were needed before the classes reached their learning plateau.2 Some of the results (the learning curves for fixed target trainer readings) are shown in Figure 3. The two curves represent at about the end of the seventh week, which means that the number of trials to reach the plateau agreed very closely with that found in the heightfinder study. Data on the rate of learning to range on aerial targets after training on stereoscopic , CLASS 8, N-87 FORT LAUDERDALE ~i 1 1 1 1 1 r i DATA FOR STEREOSCOPIC TRAINER M2 CURVE OF BEST FIT TO DATA FOR 4 CLASSES, N= 88 DATA FOR EASTMAN TRAINER, CURVE OF BEST FIT TO DATA FOR 4 CLASSES N = 46 Figure 4. Learning curve showing decrease in variability of readings on stereoscopic trainer M2.16 (Variability score scale adjusted to be roughly equal to UOE score scale in Figure 3.) trainers is shown in Figure 5. The curves are for two classes of students at NTS, Fort Lauderdale.20 Statistical analysis of these curves indicates that there is no significant improvement in performance after the first quarter or third of the training program. This confirms the results of an earlier study of heightfinder observers in which it was found that men with previous experience on the trainer needed only about 500 readings on aerial targets on the heightfinder itself to com- plete their training.1 The figure 500 corresponds very closely to a third of the training readings (1,700 total) represented in Figure 5. In summary of the present learning data, the following conclusions can be drawn. 1. Students reach a plateau in trainer learn- ing in terms of the variability of their observa- tions after about 2,500 trials. 2. Students, even those previously untrained on trainers, need no more than two weeks of training in fixed target ranging on the range- finder or heightfinder before they reach a pla- teau. 3. Following practice on a trainer, some 500 Figure 3. Learning curves showing decrease in variability of fixed target readings on stereo- scopic trainers.2 (Scale of error is expanded in comparison with Figures 1 and 2.) the records of men trained on M2 trainers and of those trained on Eastman trainers. Both curves show a learning plateau after about 21/2 weeks. Learning curves for moving target readings made by the same men parallel the present curves. The M2 trainer men took about 2,700 trials to reach a plateau, and the East- man trainer men about 3,200 trials. When the rangefinder training program was set up at NTS, Fort Lauderdale, a similar analysis of the trainer learning curves was made for 87 students in one class.16 The records for these men are summarized in Figure 4. The 8 weeks of training represented in the figure combine training on fixed targets and on moving targets, as indicated at the bottom of the graph. Throughout the training period, a total of 2,900 readings was made by each man. The learning plateau for this class was reached 46 TRAINING PERSONNEL IN RANGE DETERMINATION trials of aerial target readings bring varia- bility close to the learning plateau. It will be noted, however, that these data leave some important questions about range- finder learning unanswered. In the first place, all the records presented here are in terms of the absolute sense. As a matter of fact, the trainer validation study itself used variability scores as the validating criterion. If one were to insist on completely refined experimentation, the validation study should be repeated to de- termine how much practice on trainers can be substituted for practice on rangefinders or heightfinders in a program which has accuracy of readings as its primary objective. In the second place, the value of overtrain- ing on the trainer (or on the rangefinder) is not known. Even though a student reaches his learning plateau on the trainer in some 2,500 trials, it is possible that the skills which he transfers to the rangefinder or heightfinder continue to develop through 3,500 or 4,500 trainer trials; these extra trials would be called 1.000 or 2,000 trials of overlearning. Con- versely, it is possible that students profit only from their first 1,000 trials on the trainer even though they have not reached their trainer learning plateau in this time. Thus, if the stereoscopic method of ranging continues, there is still need for further re- search. 4-2-6 Proficiency Data Some information on the proficiency of stereoscopic operators can be drawn from the learning data presented in this section. Performance on the stereoscopic trainers is far superior to performance on either the rangefinder or heightfinder. This appears to be due to the fact that the target is always clearly seen in the trainer without atmospheric dis- turbances to trouble the observer and to the fact that instrument variance is practically absent in the case of the trainer. The latter factor is largely unknown, particularly in aerial target readings on the rangefinder or height- finder. The mean deviation of the stereoscopic set- tings of the average student on a trainer should approach 0.5 to 0.7 UOE after training in fixed target readings and should approach 0.8 to 1.0 UOE after training on moving target read- ings with tracking errors. The range in these figures allows for differences between the M2 CLASS 7, N= 78 — CLASS 8, N -8 2 FORT LAUDERDALE Figure 5. Learning curves showing decrease in variability of aerial target range readings on the rangefinder. The classes had had 8 weeks of work on stereoscopic trainers and had surface target and fixed target training paralleling this aerial target work. The men made about 100 aerial range readings per day, or a total of about 1,700 per man. The variability score scale has been adjusted to be roughly equal to the UOE score scale in Figures 1 and 2. variability of range or height observation, not in terms of absolute accuracy. Although there is a good correlation between the variability and accuracy of readings of men who have been given many hundreds of trials on the heightfinder, it is not established that 500 trials of aerial target reading are sufficient to train men in reliable calibration, i.e., in getting readings which are satisfactorily accurate in TRAINING IN STADIAMETRIC RANGING 47 trainer and the Eastman type trainers. The mean deviation of the heightfinder stu- dents’ aerial target height readings is about 2.5 UOE and of range readings is probably about 4.0 (estimated from data in Chapter 22). In general, student performance on the range- finders at NTS, Fort Lauderdale (mean devia- tion of about 5 UOE, see Figure 5) was poorer than would have been predicted from these Army data. This difference was attributable in part to poor instrument maintenance and to inconvenient and mechanically irregular track- ing gear on the rangefinders in use. 43 TRAINING IN STADIAMETRIC RANGING Stadiametric ranging is the method of rang- ing in which range is determined from the angular size of the target. In this method, the target must first be recognized as a plane with a known wingspan or as a tank of a known width or length. Its range is then inferred from its size in the visual field of a telescope or sight. Size judgments are made with reference to the known dimensions of the telescope or sight reticle pattern. To facilitate ranging by this method, many telescope reticles contain gradu- ated mil scales. A target which is 1 yard wide is at 1,000 yards range if it fills a space of 1 mil on the reticle. An attacking airplane with a wingspan of 12 yards is at a range of 2,000 yards when the wing-tip to wing-tip silhouette fills 6 mils of the reticle pattern and is at 1,000 yards when it fills 12 mils of the pattern. For field artillery operation this is not a difficult ranging method in view of the fact that the target is generally stationary or moving at a very slow rate. Measuring the angular size of the target is accomplished by adjusting the position of the telescope or sight until the reticle crosses the target in some convenient way. For antiaircraft firing, however, stadia- metric ranging is more difficult. There are at least three reasons why this is so. (1) The man who must do the ranging is often a tracker and hence he has more than one job to do. (2) The target moves around in the visual field according to the tracking errors which are made. (3) The gunsight reticle, chosen for its usefulness in tracking, is not necessarily grad- uated or designed for ease of ranging. The shifting target pattern must therefore be judged relative to some dimension of the reticle: one-half, two-thirds, or some other fraction of an area or distance represented in the reticle. A typical example is found in anti- aircraft gunsights which use a reticle pattern with a 15-mil or 25-mil diameter circle. At ranges where gunners should open fire, targets may subtend angles of only 7 to 10 mils. Thus, to the other sources of error in stadiametric ranging, there is often added the observer’s in- ability to estimate fractional parts of reticle spaces with accuracy and consistency. 4 31 Preparation of Training Manual Material It had generally been assumed that a gunner who was viewing the target through a tracking telescope could make a better judgment of when to open fire if he determined range by stadia- metric methods than if he made a free, unaided estimate. On the basis of this assumption, a number of manuals for gunners and director teams had included diagrams to show how stadiametric range estimates could be made in terms of the relative size of target and reticle. Typical manuals which had carried this infor- mation included the following Navy publica- tions : Gunners’ Bulletin No. 2, Use of the Gun- sight Mark 14, Operating Instruction for Use of the Gun Director Mark 51, Model 3, and Op- erating Instructions: Gun Director Mark 52. For purposes of simplifying the stadiametric task for the gunner or tracker, enemy planes were classified into three groups: those with wingspans approximating 36 feet, those with wingspans approximating 50 feet, and those with wingspans approximating 75 feet. Most single-engined fighter planes, carrier-based bombers, and two-engined bombers fall respec- tively into these three classes. Classification of targets could be simplified to this degree be- cause wingspan variation within these three plane types was small in relation to errors in 48 TRAINING PERSONNEL IN RANGE DETERMINATION A SINGLE ENGINE FIGHTER LOOKS LIKE THIS A CARRIER BASED BOMBER OR TWIN ENGINE FIGHTER LOOKS LIKE THIS A TWIN ENGINE BOMBER LOOKS LIKE THIS Figure 6. Appearance of aerial targets at different ranges as seen through the gunsight Mark 15 TRAINING IN STADIAMETRIC RANGING 49 stadiametric observation. Once the tracker or gunner identified the target and its class, his ranging job became one of observing when the target filled a specified fraction of the reticle. Diagrams illustrating how the three classes of target appeared at different ranges through the gunsight Mark 15 are shown in Figure 6, taken from Operating Instructions: Gun Di- rector Mark 52. 4 5 2 Proficiency of Ranging with the Reticle in the Gunsight Mark 14 In connection with a study of procedures for operating the gunsight Mark 14 on the 20 mm gun, Project N-105 compared the pro- ficiency of free range estimation and stadia- metric estimation using the reticle pattern in the gunsight. Groups of 18 men each were trained in stadiametric observations and were then tested to determine how well they could judge when the target was at 1,700 yards, i.e., when they would open fire. At this range, the target wing- span filled about one-third the diameter of the smaller of the reticle circles in the gunsight. The judgments made by each group on each trial were assembled into separate distribu- tions. The means of these distributions devi- ated 229 yards on the average from the required 1,700 yard range. The average stand- ard deviation of the distributions (18 judg- ments each) was 411 yards. 4.3.3 Methods 0£ Training Men in Ranging with the Mark 14 Reticle In an experiment to determine the relative efficiency of different methods of training gun- ners in stadiametric ranging with the gunsight Mark 14, Project N-105 compared four meth- ods of training. These were method 1, a 10- minute period of indoctrination or instruction, with the aid of wall charts, in the use of the reticle in the estimation of target range; method 2, indoctrination as in method 1 plus two 1-hour periods of estimation and training on the mirror range estimation trainer, device 5C-4; method 3, indoctrination as in method 1 plus about ten training trials on the tiring line; method 4, indoctrination and training on the 5C-4 as in method 2 plus about ten training trials on the firing line using the actual gun- sight and a real target. One day after training, each group was tested for its accuracy in stadiametric range determination using the gunsight Mark 14 on the firing line. The mirror range estimation trainer, device 5C-4, is an instrument designed by the Special Devices Division, Bureau of Aeronautics. It permits eight men to observe simultaneously the approach or withdrawal of an airplane model and to estimate the portion of an illumi- nated reticle pattern which it fills. On instruc- tional trials, the apparent range of the target, in terms of the visual angle which the model subtends, is announced at intervals and the men observe target size relative to reticle size. On test or drill trials, the target is brought in from long range and men record (by marking a moving paper disk) when they believe that the target has reached the desired angular size or opening fire range. Upward of 50 drill trials comprised the two-day experimental training program on this instrument. In training on the firing line, the men tracked an approaching target with the gunsight Mark 14 mounted on a 20 mm gun. On instructional trials, an instructor announced radar-measured ranges to the target at 5,000, 4,000, 2,500, 1,700, and 1,000 yards. Three such trials were interspersed with drill trials in which the men indicated when they thought the plane was at 1.700 yards and were advised of their errors. The procedure for having the men indicate the range at which they would open fire was as follows: The instructor who had access to radar-measured ranges called successive let- ters of the alphabet for each 200-yard point on the target’s incoming run. Thus at 4,000 yards he might call “A,” at 3,800 “B,” at 3,600 yards “C,” etc. Each gunner remembered the letter which was called at the time or just prior to the time at which he thought the range was 1.700 yards. Later he was told which the cor- rect letter was. In this way he knew whether his estimate was short, long, or just right. Since the first letter called varied from trial to trial 50 TRAINING PERSONNEL IN RANGE DETERMINATION and was called when the target was at varied ranges, the correct letter was different on each run. Seven drill trials plus three trials of an- nounced ranges completed the program of fir- ing line training. The ranging test which was given to all groups of men consisted of six trials similar to the firing line drill trials described above, except that information on estimation accuracy was withheld. Three runs for each man were with the target flying at an altitude of 2,000 feet, and the remaining runs were with the target making a simulated torpedo attack. The relative performance of the four training groups is shown in Table 1. From these results it is concluded that firing line training is, at the moment, the most satis- factory method of training men in stadiametric ranging with the gunsight Mark 14. The de- vice 5C-4 is useful only in those situations where it is impossible or impractical to train on the firing line in the real tracking situation. 44 TRAINING IN RANGE ESTIMATION Range is obtained by estimate in the casualty operation of most fire-control systems and in the standard mode of operation for a number of specific antiaircraft and tank fire-control Table 1. The effectiveness of different methods of training men the gunsight Mark 14. in stadiametric ranging with the reticle of Experimental group Number of men in group Type of target run Number of observations by the whole group Mean target range in yards when it was judged to be 1,700 yards Per cent of men whose mean judgment was within 200 yards of 1,700 yards I Indoctrination only 82 Torpedo Med. 246 2,187 23 alt. All 156 2,501 17 402 2,309 21 II Indoctrination plus 5C-4 training Torpedo Med. 246 2,080 39 for 2 hours 82 alt. All 156 2,364 20 402 2,191 32 III Indoctrination plus firing line Torpedo Med. 141 1,710 52 training for 45 minutes 47* alt. All 141 1,918 44 282 1,814 48 IV Indoctrination plus 5C-4 plus Torpedo Med. 93 1,775 43 firing line training for 45 min- alt. All 93 1,799 56 utes 31f 186 1,787 49 * Drawn from Group I. t Drawn from Group II. An inspection of the data in the last two columns of Table 1 shows that method 2 (train- ing on the device 5C-4 plus indoctrination) was more effective than method 1 (10-minute in- doctrination lecture). It tended to bring the mean point of range judgment closer to 1,700 yards and increased the number of men whose mean judgment was within 200 yards of 1,700. On the other hand, 5C-4 training was less effec- tive than a short period of ranging on the firing line. Furthermore, since method 4, which included both 5C-4 training and firing line training, did not excel method 3, it appeared that training on the 5C-4 did not add to the skill obtained from firing line training. systems. In the interest of either application it is appropriate to make experimental inquiry into the adequacy of free, unaided range esti- mation and of various training methods. The wartime need for studying range estima- tion for tank gunnery was recognized somewhat earlier than the need for the experimental in- vestigation of the accuracy of range estimation for antiaircraft firing. This was probably due to the early importance of tank warfare as well as to the possibility open to many AA batteries of getting range data from a nearby unit by telephone. Among those research agencies which studied range estimation for ground targets and tank objectives were various TRAINING IN RANGE ESTIMATION 51 British research groups,31’ 35>37 Division 7 of NDRC,11 and the Princeton Branch of the Frankford Arsenal Fire Control Design Divi- sion.30-30 Projects N-105 and N-lll of the Ap- plied Psychology Panel later investigated the accuracy of range estimation for aerial targets. Range estimation in antiaircraft firing is needed in situations of two types: for opening fire with automatic weapons and for setting range into directors or gunsights when they are operated for barrage or fly-through firing. The only difference between the two estimating problems is a difference in the range at which the estimates must be made. 4-41 Training for Estimation at Short Range It is common observation that 20 mm gunners open fire too soon; that is, they under- estimate the range of their targets. Experi- mentally, it turns out that this is true whether the men are actually at their guns or whether they are simply standing by for the express purpose of estimating ranges.19 Thus the tend- ency, at short ranges at least, to underesti- mate the range of rapidly approaching aerial targets is a very real one. It may be exaggerated by, but it is not caused entirely by, the stress and emotion of combat. It is a wasteful tendency, but, what is more, it often leaves a gunner with an empty magazine by the time the target gets within good hitting range. It is not difficult to demonstrate that this tendency to underestimate the range of attack- ing planes can be overcome by a very short training program.19 In fact, no more than three trials may be required in order to bring a group of gunners to the point where their average estimate of the opening fire range is in error by a very small amount. The training procedure used in the experi- ment was this: The plane flying on the special training mission, a TBF, was instructed to make simulated torpedo runs on the firing line where the gunners were stationed. The men had a clear, unobstructed view of the target coming in from over the ocean at an altitude of between 50 and 150 feet. During the series of three training trials, an instructor who had access to radar-measured ranges called the range of the target at 500-yard intervals from 4,000 yards in to 500 yards. Test trials to determine how accurately the men in the group could estimate a 1,500-yard opening-fire range were carried out before and after the training trials. The method of testing was the same as that described in Section 4.3.8 above, except that the men viewed the target directly, not through a gunsight. The instructor called (over a PA system) successive letters of the alphabet at 200-yard intervals during the target’s approach. The men recorded the letter which was called or had just been called when they judged the target to be at 1,500 yards range. The judgments were scored later by the research group. Thus, throughout the experi- ment the men were ignorant of how well or how poorly they were doing. A group of 151 subjects participated. Their pre-training test performance and post-train- ing test performance are summarized in Table 2. The table shows a significant increase in the number of judgments which are close to the desired 1,500-yard opening range (last col- umn). This increase results from the shift of the mean judgment for the group toward 1,500 yards (middle column of table) and an attend- ant reduction in the variation between the judgments of different men. The short program of training was successful. Table 2. Improvement of range estimation as the result of three instruction trials on the firing line. The task: to estimate 1,500 yards range for 20 mm firing. Average range Per cent of men in yards of whose estimates target when were within 15 men thought it to be 1,500 per cent of true yards range" Pre-training 1 2,374 9 trials 2 2,134 15 3 2,174 17 Post-training 1 1,612 42 trials 2 1,382 63 3 1,554 60 * The last column of the table was derived from Table IV, page 6, reference 19, and conforms with the data to be presented in Table 3. This training method, of course, does not differ significantly from that regularly in use in the fleet. The importance of the experiment 52 TRAINING PERSONNEL IN RANGE DETERMINATION was that it demonstrated how effective the method could be. In order that the facts re- vealed by the experiment would become known to the fleet and the method applied more widely, the method and test results were described in a short pamphlet, prepared jointly by Project N-105 of the Applied Psychology Panel and the Antiaircraft Training and Test Center, Dam Neck, Virginia, Instruction in Range Estima- tion on the Firing Line,38 Project N-105 also evaluated the mirror range estimation trainer, device 5C-4, for use in training men in free range estimation.18 It was found that 1 or 2 hours of drill25 on the 5C-4 in no way improved the accuracy of range estimation by the experimental group. When these men were tested on the firing line they performed just as poorly as men who had had no training in range estimation whatso- ever. Thus, although the 5C-4 is useful to a degree in the training of men in stadiametric ranging (Section 4.3.3), it is of no value in free range estimation training. 4-4 2 Training for Estimation at Medium Range The accuracy of aerial target range estima- tion at medium and short ranges was studied by Project N-lll of the Applied Psychology Panel.22 Data were collected for ranges between 1,500 and 8,000 yards, the range in which esti- mates are needed for target designation and in the use of directors operating without ranging systems. The experiment sought to determine and compare the accuracy of estimates made by trained and untrained men. Although the num- ber of subjects in the experiment was not very large, the results for several parts of the ex- periment were sufficiently consistent to indi- cate that the data are dependable. Four groups of men, totaling 18 in all, participated. Training Methods The training which each of the four groups had varied considerably. Group A consisted of five fire controlmen. They were tested before and after a 1-month period of training in the operation of the gun director Mark 52. Work with this director in- volved some estimation of ranges, usually fol- lowed by a radar operator’s report of the actual range to the target. Although this range esti- mation training was informal and a by-product of work with the director, the men improved significantly in their ability to estimate target ranges. Group B consisted of three fire controlmen who participated in a training study. Their initial estimating ability was measured in a test extending over 3 days before training began. Training sessions ran for 6 days. The training procedure was as follows. As the men observed a designated target, the radar op- erator gave instructions by loudspeaker, “Ready. . . . Estimate.” Each observer recorded his estimate on a prepared record form and then looked up to watch the plane as the radar operator announced the range of the target when its range was 200 yards and then 400 yards different from range at the time of esti- mation. Three estimates at well-spaced ranges were usually made on a single target. About 150 training observations were included in the 6-day period. Then a 2-day post-training test was conducted. Although the men were poorly motivated during the experiment, their rec- ords showed considerable improvement in ac- curacy of estimation as a result of the training. Group C consisted of seven officers, three of whom had had no sea experience at all and four of whom had had sea experience in fire control or gunnery. These men were given no training as part of the experiment, but they were given a 2-day test to compare the abilities of experi- enced and inexperienced groups. The group with sea experience, and with such incidental or formal range estimation training as that provides, were significantly better estimators than the non-sea-experience group. Group D included three research men who were drilled in range estimation. Training began with a procedure identical to that for Group B, with the men making three estimates on each target, but later shifted to a routine in which the men made a single estimate on each target and were then advised by loud- speaker of the range of the target at every TRAINING IN RANGE ESTIMATION 53 500-yard point as long as it was practical to follow the target. Training continued for 10 days, spaced over a four-week period, during which time each man made about 250 estimates followed by range information. Each man in the group became significantly less variable in his judgments and the accuracy of the group became the best of any of the groups studied. Testing Procedure for Estimation Experiments The observers stood in an enclosure 15 feet square and 6 feet 4 inches high. The walls of the enclosure obscured all land objects, so that the aerial targets were always seen against a free expanse of sky. Each observer recorded his range estimates on a record form which listed ranges from 800 yards to 10,000 yards. Ranges between 800 yards and 2,000 yards were listed by 100-yard intervals, ranges be- tween 2,000 and 5,000 yards by 200-yard inter- vals, and ranges between 5,000 and 10,000 yards by 500-yard intervals. These recording intervals represented range steps which varied between 5 per cent and 10 per cent of range. Previous records of range estimation for ground targets30,31,37 indicated that these recording intervals would be below the ob- servers’ limit of accuracy for aerial target range estimation. Estimates were made on “targets of oppor- tunity”—targets on any course, at any alti- tude, of any size, and observed in weather varying from clear to hazy. No data were secured on days when visibility was below 5,000 yards. Although the record form indicated ranges from 800 to 10,000 yards, estimates were never called for when a target was under 1,500 yards or over 8,000 yards. Since the observers were not advised of this, the response scale was “open” at both ends. Target ranges were measured with a radar Mark 26 mounted on a gun director Mark 52. The experimenter selected a target, pointed to it, and called its bearing. When the director and radar were “on target,” and when all observers reported that they could see it, a series of three estimates on that target was begun. To signal each estimate, the radar operator called, “Ready. . . . Estimate,” over a portable loud- speaker system. Each signal was called when the target range was exactly equal to one of the ranges listed on the record form. After the first signal, each observer put a number 1 oppo- site the listed range corresponding to his esti- mate. A second and a third estimate were called for at later points in the target’s course when range had changed. Each man marked his new estimates on the record form as number 2 and number 3. Twelve runs, three estimates per run, generally constituted a day’s test. The radar operator so varied the ranges at which estimates were called for that at the end of each day’s test there were about the same num- ber of observations in each of five predeter- mined and logarithmically equal range zones (1,500 to 2,000, 2,100 to 2,900, 3,000 to 4,000, 4,200 to 5,000, 6,000 to 8,000 yards). Data Analysis The data were analyzed on logarithmic charts. Figure 7 illustrates the plotting pro- -i 1 r ESTIMATES ON INCOMING TARGETS ESTIMATES ON OUTGOING TARGETS OBSERVER J 5-18-44 STANDARD DEVIATION OF ESTIMATE ERRORS = 2.2 AVERAGE MISS = 2.0 PER CENT OF ERRORS WITHIN 15% OF TRUE RANGE =43%. Figure 7. Estimates of aerial target ranges. cedure. Target ranges are given on the logarith- mic horizontal axis, range estimates on the logarithmic vertical axis. The central diagonal line represents estimates which are equal to 54 TRAINING PERSONNEL IN RANGE DETERMINATION true range. The two diagonals either side of the correct estimate line are error lines repre- senting a logarithmic error of 0.07. These side diagonals mark a band which includes all esti- mates from those which are 15 per cent shorter than true range to those which are 17 per cent longer than true range. On a log-log plot, the pattern of the plotted estimates provides a direct clue to the way in which estimate errors change with range. If the pattern of points is equally wide at all ranges, it may be concluded that variable errors of estimate increase linearly with range. If the pattern of points departs from the central diagonals, the observer has a bias toward un- derestimation or overestimation. When the scatter of points in Figure 7 is examined, it is seen to be of about equal width at all ranges and to reveal little general bias. The observa- tions made on incoming targets and those made on outgoing targets showed a shift in bias, however, the circles on the average being lower on the chart than the corresponding crosses. Once the estimates for a given man on a given day had been plotted, his errors from true range were tallied to form a frequency distribution. This tally was made directly from the graph using a logarithmic collection inter- val corresponding to a 10 per cent range decre- ment (that is, a logarithmic step of 0.046). Such a logarithmic unit is convenient for measuring systematic estimate errors and variable estimate errors since it gives the same weight to equivalent relative errors at all ranges. To interpret these logarithmic error scores for any observer, it is sufficiently accu- rate to multiply the score by ten and consider the result as an error measured in per cent of true range. Three measures of observer performance were used, two measures of error and one measure of successful estimation. 1. Standard deviation of estimate errors. After a frequency distribution of estimate er- rors (in log steps) had been tallied from the log-log charts, the standard deviation of that distribution was determined. For the estimates plotted in Figure 7, the standard deviation is 2.2 (interpreted as about 22 per cent of true range). 2. Average miss, or average error from true range. From the frequency distribution of esti- mate errors (in log steps), the average error, neglecting algebraic sign, was determined. For the estimates plotted in Figure 7, the average miss was 2.0 (interpreted as about 20 per cent of true range). 3. Per cent of estimates within 15 per cent of true range. This was based on a count of the number of estimates falling within the central diagonal zone on the log-log chart—the esti- mates which fell within 1.5 logarithmic error units of true range. Forty-three per cent of the estimates plotted in Figure 7 fell within the 15 per cent zone. Results The results for the four groups of subjects are summarized in Table 3. All the observations by the men in any one of the groups were massed in a single distribution for purposes of this final summary. The data in the table indi- cate what can be said about the accuracy of single observations taking into account the variability of estimates for any one man, the variability of estimates from man to man, the variability introduced by changes in weather and visibility conditions from day to day, and the variability introduced by changes in target size and target course. It will be observed from an inspection of the table that the per cent of estimates which fell within 15 per cent of true range for each of the trained groups was about the same. These figures, in excess of 40 per cent, are smaller than those presented in Table 2, Section 4.4.1 above, for the case of estimating the range to a target on a torpedo run. Although the train- ing was very brief in the case of the latter study, the excellence of the estimates in terms of the per cent of individual judgments within 15 per cent of true range is to be accounted for most probably in terms of the simplicity and constancy of the observation situation from trial to trial during training and test. The combined results of the two studies suggest that gunners who are trained especially to estimate opening range for 20 mm firing can be ex- pected to cluster their observations closer (on a per cent of range basis) to the desired range TRAINING IN RANGE ESTIMATION 55 than director operators who must estimate ranges over a wide variety of conditions and ranges. Table 3 shows that practice in aerial target range estimation increases the per cent of observations within 15 per cent of true range from about 25 per cent to about 40 or 45 per cent. At the same time the average error of a single range estimate drops from about 33 per cent of true range to about 22 per cent of true range. These values represent general averages for the four groups of subjects. All forms of men should participate actively by recording their observations. Their performance should be scored promptly in some way, so that the men learn quickly of their errors. Until other data are forthcoming, the data cited in this chapter can be used as standards against which to compare the scored performance of Service groups in training. The tendency to underestimation which is so prevalent within 20 mm gun range was not observed in estimates at longer range.22 This may mean that estimates are seriously influ- Table 3. Training data : aerial target range estimation at medium ranges. Total number of estimates made by the group Standard devia- tion of estimate errors Average miss or error from true range Per cent of esti- mates within 15% of true range Group A: 5 fire controlmen 1-day test before work on Mark 52 150 3.3* 3.6 25 1-day test after 1 month of work with Mark 52 493 2.4 2.3 41 Group B : 3 fire controlmen 3-day test before range estimation train- ing 400 V < i%t 3.4 3.0 30 2-day test after 6 days of range estima- tion training 153 2.9 2.5 41 Group C: 7 officers 3 officers without sea experience (2-day test) 185 p — 2%f 3.2 3.2 23 4 officers with fire-control and gunnery experience (2-day test) 186 2.6 2.1 42 Group D: 3 laboratory men Data for 6 days after 10 days of train- ing (about 250 reinforcements) 400 P < l%t 2.1 1.9 45 * Multiply by 10 in order to interpret as per cent of t Probability that difference between the above pair range. See text, of figures could have been due to chance factors alone. training represented in the experiment were effective, but the best range estimation per- formance was obtained with the most interested group of subjects, Group D, which was given the most formal and longest training. 443 General Conclusions on Range Estimation Training Relatively short periods of training and drill in situations resembling combat situations and involving actual airplane targets can be very effective in reducing range estimation errors. In such training, the gunners or fire control- enced by the observer’s “set” and that “to open fire” establishes a different set than “to put range in a director.” More research on this problem is in order. Relatively frequent refresher drills, perhaps one every 2 or 3 weeks, may be required to keep men up to par in their estimating skill. Not many data were gathered on the subject of the retention of range estimation skill in the course of the wartime studies, so this is a field which still requires careful investigation. The only pertinent data which can be cited are the following. In the estimation of opening range for 20 mm operation, there is some decre- ment in estimation accuracy 11 days after 56 TRAINING PERSONNEL IN RANGE DETERMINATION training, but there is still a significant amount of skill retained.20 In the experiment on medium range estimation, each of the Group D men was just as consistent in his judgments 60 days after training as he had been at the end of training, but he was no longer as accurate. Biases appeared in their observations. This corresponds to a loss of calibration, if you will. Short and frequent periods of drill on friendly targets would overcome this trend toward in- accuracy by permitting each man to reestablish his calibration and bring his estimation scale once more in line with the scale of true ranges. The foregoing data on the estimation of range to aerial targets are in agreement with the general results on the estimation of range to ground targets for artillery and tank firing. Average range estimation errors increase in an essentially linear way with range. This finding, confirmed for aerial target range estimation, parallels the data on sensory discrimination which are subsumed under the Weber-Fechner law of discrimination—that the ratio of obser- vation error to the magnitude observed is a constant. Thus, unaided visual range estima- tion resembles, in its quantitative character- istics, other more completely analyzed and more commonly studied forms of human perception. Chapter 5 TRAINING THE B-29 GUNNER Charles W. Braya SUMMARY IN CONNECTION with other research on B-29 gunnery, Project AC-94 of the Applied Psy- chology Panel studied the scoring of gun camera film, a checksight for B-29 gunners, and the effects of coaching in a training program. A new scoring method was developed which required less than one-third the number of man-hours of the standard Army method. The checksight provided unreliable scoring but probably motivated the men to better perform- ance. Coaching had less effect than expected on the basis of general psychological experi- ence. The data of the coaching experiment indicated that gunners adopted a standard pat- tern of ranging, tending to over-range at the beginning and to under-range at the end of an attack. 51 INTRODUCTION In connection with a program of research on psychological problems in the design and operation of the B-29 gunsight as described in Chapter 20, the training of the B-29 gunner was studied by Project AC-94 of the Applied Psychology Panel. Research problems of design and operation can be solved only by the de- velopment of adequate methods of measure- ment of gunner performance and by studies of the ability of the average gunner to learn to use his equipment. Thus research on design furnished by-products in the form of contri- butions to training problems. The B-29 gunner had to track and to range simultaneously. In all studies conducted by Project AC-94, the gunsight (Figure 1, Chap- ter 20) required position tracking through an optical sight; a spot of light in the center of the field of view of the sight was held on the target. Around this spot were a varying num- ber of illuminated diamond-shaped dots out- lining a circle. The diameter of the circle was controlled by the gunner, giving a measure of target range when the apparent distance from wing tip to wing tip just equaled the diameter (adjustment for variation in target size was made in advance of an actual attack). The gunner’s task was difficult and complex. He was required to track skillfully (the prob- lems of tracking without simultaneous ranging are considered in Chapter 3). Accurate ranging required that the wing tips be framed by the reticle. Inadequate tracking required a change in the method of ranging from the relatively simple process of framing to the more difficult process of comparison of two dimensions. When the target approached at an angle other than head-on the gunner had to correct for the effects of perspective on wing length. In addi- tion, since the reticle circle was defined only by eight to twelve discontinuous dots, it was seen by some gunners as a polygon rather than a circle. Also, the central dot, which should have helped to define a diameter, was frequently disregarded as the gunner’s attention shifted from tracking to ranging. Thus the gunner sometimes chose a chord rather than a diameter as the critical dimension. Finally, the attacks were very short in duration because of the high speed of the B-29 airplane. In view of the difficulty of the task, the suc- cess of the B-29 gunners is worthy of remark. The data given below and in Chapter 20 show that he could track with an accuracy approach- ing 5 mils and that the average error of rang- ing could be reduced to about 10 per cent of range. These figures were by no means typical of standard Army results in B-29 gunnery. They were achieved after special training on a few types of attack and only on the ground. Nevertheless they were achieved by standard Army personnel. The objective of the research of the Applied Psychology Panel, in collabora- tion with the Research Division, Laredo Army Air Field, was to attain similar results under all conditions. Contributions were made by Project AC-94 a This chapter is based on the work of Project AC-94. 57 TRAINING THE B-29 GUNNER 58 in three phases of the training program: (1) the measurement of gunner performance, (2) training methods, and (8) the development of synthetic trainers. Because the last topic is so closely related to the program of evaluation of B-29 gunnery equipment, it is discussed in Chapter 20 rather than in this chapter. Certain technical questions relative to the measurement of gunner performance are also treated in that chapter. In the discussion which follows it is assumed that the reader is familiar with Chap- ter 3, Training Trackers for Antiaircraft Fire Control. 52 THE MEASUREMENT OF GUNNER PERFORMANCE A basic need in any training program is to measure the performance of the trainee. The instructor needs to know men’s scores to de- termine which need coaching and which have achieved a level satisfactory for “graduation.” If a score is furnished the trainee, he will com- pete with others and with his own past per- formance. For both instructor and trainee there is need for a valid, reliable, and informative score that will be available immediately or as soon after the performance as possible. In the fall of 1944 the training of B-29 gun- ners was accomplished in three ways: on the ground through use of synthetic targets, in the air through use of simulated attacks on a bomber by pursuit planes, or in a mixed situa- tion by simulated attacks of a real plane on a ground mock-up of a bomber. In all phases a contribution was made to the measurement of performance. The development of scoring with synthetic targets is described in Chapter 20. 5 21 Scoring Gun Camera Film In air-to-air and mixed practice, the gun camera was in standard use. The gun camera, attached to the sight and photographing the sight reticle and target, furnished a film rec- ord of gunner performance. At the completion of practice the film was developed and errors were measured. Frequently the film was also shown to the gunner to indicate the quality of his performance. Since thousands of gunners were under training, use of the gun camera was a major undertaking. Photographic de- velopment, alone, was a difficult problem. Scor- ing required over 6 man hours for about 6 minutes of “firing.” These 6 minutes included about 50 representative attacks. Method of Scoring The standard method of scoring was to measure every tenth frame of the projected film. The size of errors in tracking and framing was stated in terms of millimeters. This unit was relatively meaningless to the gunner. Thus the method provided less information than was desirable, it was arduous, and the information was given the gunner long after the perform- ance. Nevertheless, when time permitted its use, it provided the only quantitative informa- tion received by the gunner. The project simplified1 the scoring of gun camera film by an adaptation of the checksight method, per cent of observation instants on target, described in Chapter 3. The method developed for the gun camera was named the on-target method. In on-target scoring, each frame of projected film was judged to be on the target in tracking if the center dot of the reticle touched a part of the target fuselage forward of the trailing edge of the wings. It was judged on the target in ranging if the diameter of the reticle circle was within ± 5 per cent of the distance from wing tip to wing tip. The on-target score consisted of the per- centage of all film frames in which the gunner was within the tolerances for tracking and ranging simultaneously. This score seemed more meaningful to the gunner than the arbi- trary millimeter score used in the standard procedure. Normally the scorers estimated whether the film met the tolerances, but they checked them- selves by measurement in case sequences of doubtful frames occurred. A method of train- ing observers in making the estimate was de- veloped. Observers should be given refresher training from time to time to check any tendency toward a shift in the tolerances. THE MEASUREMENT OF GUNNER PERFORMANCE 59 From the same data two other scores were derived, the computer-on-target score and the hits-on-target score. The computer-on-target score was developed in order to remind the gun- ner that the B-29 computer does not instan- taneously solve the problem of integrating the input from the gunsight. After the gunsight has moved, there is a delay before the effect of the sight movement is fully realized in gun movement. The length of the delay varies but a standard delay of J/2 second was arbitrarily assumed to be accurate enough for training purposes. On this assumption the gunner had to be on target in tracking and ranging for V2 second before the computer drove the guns into proper firing position. The computer-on- target score, therefore, was calculated by sub- tracting eight frames (V2 second) from the number of frames in each sequence of on-target frames. The remaining frames in the sequence were computer-on-target frames. The sum of these expressed as a percentage of the total number of frames for the attack was the com- puter-on-target score. The hits-on-target score was based on computer-on-target frames occur- ring when the gunner pressed the trigger. The hits-on-target score was not evaluated in the work of the project, and the computer- on-target score was evaluated only for the ground mock-up. The tolerances for air-to-air firing, as will be shown below, were set too narrowly in the Panel’s studies, and computer- on-target scores were usually zero in aerial practice. Evaluation of Scoring Accuracy To evaluate the on-target scoring system a group of 20 experienced gun camera film scorers from all Air Forces training centers was assembled. Over a period of 5 days these enlisted men were trained in the use of the new method and given refresher training in the old by scoring and rescoring film for 50 attacks. A different set of 50 attacks of ground mock-up film and 40 attacks of air-to-air film were then scored twice by each method (orders of scoring counterbalanced). In Table 1 are stated the general level of ability and the variability of performance shown by the sample films used in the experiment. For these films, originally thought to be rep- resentative of gunners in operating training, the mean on-target scores were 29 per cent for the mock-up and 13 per cent for air-to-air attacks. It will be shown below that these figures give too high an estimate of the ability of the gunners at this stage of training. Since the most effective on-target scoring system gives mean scores of 50 per cent, it must be concluded that the tolerance limits within which the gunners were judged on target were set too narrowly. Nevertheless enough varia- bility existed in the sample films to warrant a preliminary correlation analysis of the relia- bility of the method. Table 1. Average performance in gun practice when scored by various methods. camera Score 50 Mock-up attacks 40 Air-to-air attacks Average at- tack scores o distri- bution Average at- tack scores o distri- bution On target 29.11% 23.48% 13.28% 15.37% Computer on target 14.11% 16.74% 1.64% 4.30% Measured tracking error 5.54 mm 3.32 mm 11.39 mm 5.40 mm Measured ranging error 12.81 mm 11.07 mm 12.99 mm 9.89 mm In Table 2 the consistency of the individual scorers is shown by the mean reliability coeffi- cient for first and second scorings by each method. Ranging error is more precisely de- termined by the measuring method of scoring, but the obtained reliability of the on-target method is adequate. The data may be expressed in another way by the statement that the average difference from first to second scoring was roughly 1 to 8 per cent for the various conditions and scores by the on-target method, and roughly 1 to 2 mm by the measuring method. A study of inter-scorer differences gave the same conclusion. Throughout all comparisons, measured framing error was the most reliable and objective score, but the on-target method was reasonably satisfactory. The correlations of the two kinds of score with measured accuracy on every frame of the film were calculated. The correlation of meas- 60 TRAINING THE B-29 GUNNER ured accuracy in ranging on every tenth frame with measured accuracy on every frame for 29 of the attacks was .999, the average difference being only 0.2 mm. For the on-target score, however, the correlation with measured ac- curacy on every frame in ranging was only .68, and 12.7 per cent of all frames were incorrectly be scored because of poor photographic quality. The results of the follow-up experiment indi- cated that the performance of these gunners was very poor. Their on-target scores are shown in Table 3. The table indicates a low level of achievement in this training which just preceded combat. The distributions are J-shaped with a preponderance (50 to 72 per cent) of zero scores. A study of the 28 gunners who had scorable film for successive sessions gave small evidence of improvement as train- ing progressed. Table 2. Mean of 20 score-rescore reliability cor- relations of gun camera film by various scoring methods. Mean r for Mean r for Score 50 mock-up 40 air-to-air attacks attacks On target .86 .84 Computer on target .81 Measured tracking error .88 .82 Measured ranging error .97 .93 * Such a high proportion of air-to-air computer-on-target scores were zero that this correlation coefficient would be meaningless. Table 3. Distribution of on-target scores in scor- able gun camera film of 250 B-29 gunners during operational training, March to July 1945. The units for table entries are single attacks. On-target score Number of attacks Session 1 Session 2 Session 3 Session 4 0 Manipulation tower 188 136 108 55 1-10 38 32 31 9 11-20 14 14 13 4 21-30 10 9 7 3 31-40 6 2 5 3 41-50 2 4 4 2 51-60 61-70 Total 2 260 0 1 198 168 76 0 209 Air-to-air 175 87 33 1-10 83 63 31 19 11-20 25 26 17 9 21-30 17 23 10 3 31-40 2 3 4 2 41-50 5 3 2 51-60 1 3 1 61-70 71-80 81-90 Total 0 0 1 343 296 152 66 judged as being on or off target as defined above. The data indicated that the scorers were generally too strict. Although the measurement method showed some superiority in reliability (for ranging) and in accuracy, it required 8.3 times as many man-hours to score the film by it as by the new on-target method. These results of the comparison of scoring methods show a very small loss of reliability when estimation replaces measurement. The lack of a greater difference between the two methods is understandable if one considers the details of the measurement method actually used on gun camera film. For tracking, the scorer judges the center of the target’s nose and measures the distance of the reticle dot from this judged center. The judgment may vary considerably. In the case of ranging, the wing tips are often poorly defined on the projected image. In addition, the time required to make measurements permits the film to heat and expand between measurements of reticle and target sizes. The on-target method was applied in a fol- low-up study4 to the film of 250 gunners who received, on the average, three sessions each of mock-up training and air-to-air training during the period March to July 1945. The number of sessions varied considerably for the individual gunners. Much of the film could not 522 A Checksight for Scoring B-29 Gunners The many defects of the gun camera were obvious. At best it provided a score or per- mitted coaching long after the completion of a particular practice session. In addition, it was very expensive, A substitute method was de- sirable. On synthetic motion picture trainers Army instructors often modified the optics of the gun- sight so that the reticle was projected on the screen. Thus the instructor could see the results AN EXPERIMENT IN TRAINING B-29 GUNNERS 61 of the gunner’s efforts and have a basis for coaching him. It was recognized that the in- structor’s judgments were relatively unreliable in this situation, but it seemed possible that the unreliability was due in considerable part to the poor definition of film targets. Hence, in the summer of 1945, an Army technician de- veloped a checksight for use on real targets. This checksight was a hand-held telescope in which reticle size was controlled by a repeater from the gunsight. Project AC-94 undertook the evaluation of the checksight.2 With it, the instructor tracked the target and observed the gunner’s ranging. At one-second intervals the instructor informed the gunner that he was “over,” “under,” or “good” in ranging. The judgments were re- corded and later compared with gun camera film. The accuracy of judgment was low. How- ever, the performance of the gunners in the experiment was very good. It seemed possible that the checksight, despite its inaccuracy, served as a motivating device. Additional studies of the device were planned, but the closing of the project prevented further re- search. 53 AN EXPERIMENT IN TRAINING B-29 GUNNERS A single, preliminary study in training methods for B-29 gunners was carried out un- der Project AC-94.4 The primary purpose of the study was to compare a training program in which the gunner’s faults and successes were analyzed and explained to him immediately after his practice with a program in which no such coaching occurred. The analysis and explanation of the gunner’s faults and successes were made possible by recording tracking, ranging, and triggering on a moving-paper kymograph. The movements of the appropriate sight controls were transmitted mechanically to the kymograph pens (Chapter 20). A modified E-14 trainer provided a motion picture target. Two groups of 12 inexperienced gunners each and a third group of 10 graduates from a basic school for B-29 gunners served as subjects. The two groups of inexperienced gunners were bal- anced for GCT score. They were also equated for original proficiency on one attack, but it was discovered later that this did not serve to equate them on all attacks. One group of inexperienced gunners was coached over the 20 days of the experiment and one was not. The graduate group was un- coached for the first 8 days and then coached until their training ceased on the fourteenth day. Coaching consisted of analyzing, with the aid of the gunner, two of six recorded attacks out of a total of 16 attacks for each man on each day of the experiment. The 16 attacks were those on the training film P-7, which gave beam attacks of fairly high speed. Every man in the experiment received a daily score on his ranging performance. This score was the mean (regardless of sign) measured ranging error at V3-seconcl intervals through- out one of the recorded attacks. Use of this error score as a motivating device distinguished the training of all gunners in the experiment from standard Army training. The results on the main experiment may be stated in terms of framing error on five of the recorded attacks. Mean performance of each of the three groups for all five attacks is shown in Figure 1. The coached group was slightly su- perior to the uncoached group. The superiority appeared throughout the experiment. The dif- ference persisted at a confidence level of .06 when the original superiority of the coached group was taken into account. Thus the dif- ference between the groups was in the direc- tion expected on the basis of other psycho- logical studies, but the magnitude of the difference was too small to be accepted with confidence. The experiment also showed that the gradu- ate gunners were no better at the beginning than the inexperienced men. Nor did the gradu- ates improve more rapidly. It is evident that standard Army training was essentially in- effective, at least as far as ranging accuracy was concerned. It is also evident from the data that training which includes a motivating score is effective in improving performance, not only for inexperienced men but for experienced men as well. An average error in ranging of around 62 TRAINING THE B-29 GUNNER 10 per cent of range is a satisfactory perform- ance relative to the 20 per cent error, which was typical of all groups at the beginning of the experiment. That the final performance level of all groups was satisfactory is also suggested by a follow- up in air-to-air firing. The study used the firing error indicator, which is an acoustic device mounted in a towed glider to provide a measure of the accuracy of actual fire on a target. Against 18 crossing-over attacks for each man the gunners of the experiment fired a total of 15,000 rounds. According to the firing error indicator 5.7 per cent of all rounds were within 2.5 yards, 7 per cent were within 2.5 to 5.0 yards, and 18.6 per cent were within 5 to 10 no question of the adequacy of the training given the men. There were no differences among the three groups of the experiment. Un- fortunately, comparable data for men trained only by Army standard methods were not se- cured before the war’s end. Perhaps the most interesting result of the ground experiment was the indication that ranging errors show systematic variation in the E-14 trainer situation. On all five attacks, there was a tendency toward overframing of the target at the beginning of the course and underframing as the target closed. The tend- ency persisted despite an overall constant error of underframing on a fast attack and of over- framing on a slow attack. The data suggest that the gunners tended to adopt a pattern of ranging which was independent of the target movement. Any further experimentation on this subject should consider at least the fol- lowing: the poor definition of motion picture targets; the relatively large size, at long ranges, of the dots defining the reticle circle; and the overall complexity of the gunner’s task which may lead him into a series of more or less auto- matic and nondiscriminating reactions (Chap- ter 20). 54 CONCLUSION The Panel’s studies of the training of B-29 gunners demonstrated that time could be saved in the scoring of gunner performance for training purposes and that improvement in training methods was possible. In addition, they suggested that systematic experimentation on training might yield some understanding of the limitations on gunner performance which the difficulty and complexity of the task impose. UNCOACHED N = 12 COACHED N = 12 GRADUATE N = 10 Figure 1. The effects of coaching on mean rang- ing error for 20 days of practice. The group of graduate gunners was uncoached for the first 8 days and coached thereafter. yards of the target. If such accuracy were attained against real targets there would be Chapter 6 NIGHT LOOKOUTS By Dael Wolfle* SUMMARY A shipboard STUDY of the performance of night lookouts on duty in an Atlantic con- voy provided specific information on the aver- age ability and the variation in ability of night lookouts to spot targets. The best lookout was able to see a ship nearly four times as far away as the poorest one. This difference was not due to differences in night vision as measured by the adaptometer. Two night vision trainers and a training exercise in the use of binoculars at night were constructed and submitted to the Navy. Assist- ance was given in the preparation of literature for the training of lookouts. 61 INTRODUCTION The lookouts, the “eyes of the ship,” have always played an important role in naval op- erations. They watch for other ships, friendly or enemy, or for lights, smoke, or debris that may indicate the presence of another ship. They help in navigation as well as in combat. In more recent years submarines, periscopes, torpedoes, and airplanes have been added to the list of things to watch for and report. In one sense the task of the lookout is a very simple one. He merely scans his sector of the sea or sky and reports what he sees. But in another sense his difficulties are fairly great. Looking steadily at a sector of the sea with nothing in sight but the sea is not conducive to a wide-awake alertness. So questions of fatigue and of motivation are added to the obvious visual aspects of the lookout’s job. Air- planes approach with such high speed that systematic and rapid scanning becomes neces- sary. At night the lookout must learn to acquire and maintain dark adaptation and must learn how to use his eyes most effectively in the dark if he is to pick up targets in time for appropri- ate action to be taken. In order to make accurate estimates or readings of the range, position angle, target angle, and relative bear- ing of a target, he must learn a number of special techniques and must learn how to use the equipment supplied him. In order to in- crease his range of vision he must learn how to adjust and use binoculars. All of these require- ments add to the difficulties of the lookout’s task and create problems of proper selection and adequate training. The Navy, well aware of these problems and aware also of the rather low prestige of the lookout’s job, a prestige which made it diffi- cult to assign good men to this job, asked the Applied Psychology Panel to make “an analysis of the duties and successes of officers and en- listed personnel in various night lookout assign- ments with reference to the specific abilities required for successful duty” and to suggest improvements in the training of night lookouts. The work was assigned by the Applied Psy- chology Panel to Project N-115, Princeton University, The major work of this project was an effort to obtain reliable, quantitative information on the performance of night look- outs under actual wartime shipboard condi- tions. In order to secure this information, arrangements were made by the Navy for two members of the project staff to conduct a study aboard a cruiser engaged in convoy duty. 62 THE PERFORMANCE OF NIGHT LOOKOUTS ABOARD SHIP At sea, in a convoy, studies2 were conducted on the ability of the ship’s regularly assigned lookouts to spot targets visually at night. The subjects for this study were the ship’s regular lookouts. They had been chosen for their duties by their respective division officers and were under the supervision of a lookout officer and an assistant lookout officer for purposes of organization and training. Most of the men were seamen 2/C and all were unrated. Con- a This chapter is primarily based on the work of the staff of Project N-115. 63 64 NIGHT LOOKOUTS siderable evidence indicated that these men were typical of those who may be found gen- erally throughout the fleet performing lookout duty. 6,2 1 Collection of Data Observations were, of course, made only dur- ing the hours of darkness. One of the investi- gators stood on the open deck at the level of the forward-sky-lookouts. He made frequent quali- tative and quantitative records of weather conditions and visibility. The second investi- gator was stationed in the combat information center [CIC]. Both were in direct communica- tion with all lookouts and with each other on the regular lookout [JL] circuit of the ship’s sound-powered telephone system. The investigator in CIC called each lookout in turn on the JL phones, asked for the bearing of the most distant ship he could see in his sector, recorded the bearing of the target as reported by the lookout, and obtained the range of that target from the duty radar operator. At the same time he recorded the name and station of the lookout, the date, and the time. Weather, sky, and sea brightness conditions were obtained by phone from the investigator on the open deck and also recorded. The type of ship sighted (transport, tanker, destroyer escort, etc.) was obtained later from a chart of the convoy. At the beginning of each watch men came directly from compartments lighted only by dim red light. Furthermore, testing did not begin until the men had been on watch for 15 minutes or more. All men were therefore dark- adapted before being asked to make any report. All tests were made when the lookout was using standard Navy 7 X 50 binoculars. With few exceptions all bearings were read from alidades. When the alidade could not be read, the bearing was estimated by the lookout. The investigator in CIC then got from radar the range of the nearest ship in the vicinity of the estimated bearing. On most nights it was possible to secure reports during only one watch. Moonlight and the short hours of summer darkness prevented a longer observation period. The cruiser was usually in the first or last line of the convoy formation. In either case all visible targets were therefore within a 180-degree sector. Conse- quently, it was usually possible to secure re- ports from only part of the 14-man watch. On some nights the destroyer escort [DE] screen could be seen, making it possible to secure reports from all lookouts on duty. The raw data of this experiment consisted of bearings and ranges of the most distant ships visible to lookouts when they were asked for a report. In their original form, these data provided no means of comparing lookouts di- rectly. The size and distance of the target ship, the angle to the target, the illumination, and the height of the lookout all varied. It was therefore necessary to introduce a considerable number of corrections to make the reports comparable. Two main types of correction were necessary, those for purely geometrical fea- tures of the situation and those for differences in illumination. Geometrical Corrections The visible area of a target is reduced when the ship is foreshortened as a result of the angle at which it is sighted (target angle). The area of the image thrown on the observer’s eye varies inversely as the square of the range at which the object is seen. And the actual size of the ship is a variable that obviously has to be taken into account. These three correc- tions could be made easily and with a reason- able degree of accuracy. Target Angle. The corrections for the fore- shortening effect were read from Table 1 which provided an approximation to the sines of the target angles. Seven discrete steps were used instead of a continuous function. This table gives the visible percentage of a ship’s length with satisfactory accuracy for present pur- poses. In making use of Table 1, the target angles for most ships were found directly from the bearing, since their course was always that of the ship from which the sighting took place. The DE’s, however, constantly changed their course with respect to that of the convoy, so that no simple relationship existed between THE PERFORMANCE OF NIGHT LOOKOUTS ABOARD SHIP 65 angle of sighting and target angle. It was neces- sary, as a consequence, to choose some arbitrary figure, and since 40 degrees was assumed to be on the safe side, the “visible length” of a DE was always taken to be 200 feet except for those cruising directly forward or astern of the look- out. Since these latter also changed course somewhat, they had to be rated at a visibility above what would be offered by their beam alone. A standard of 90 feet was used in all the calculations involving these ships. Range to Target. Corrections for range were made by using a distance of 1,000 yards as a point of reference and dividing all areas by the quantity (R/1,000)2, where R is the actual range in yards obtained from the radar op- erator at the time of the report. Thus all rec- feet; cargo ships, height 30 feet, length 450 feet; tankers, height 20 feet, length 450 feet; and DE’s, height 30 feet, length 300 feet. Corrections for Illumination Since visual acuity varies with illumination, it was necessary to correct all reports in terms of the light available for seeing. The final value of “equivalent square feet” therefore involved not only a standard range but also a standard brightness. The standard used in these calcula- tions was 0.05 foot-lambert, which is about the brightness of the sky in bright moonlight and is in the neighborhood of the brightest illumi- nation recorded on the cruise. It is difficult to arrive at a precise method of correcting for illumination, since there are no perfectly satisfactory tables showing the relationship between small changes in bright- ness and human visual acuity. Two sources of information were, however, available. Very accurate data secured by Konig almost 50 years ago give the relationship for light varying from very low values to very high ones, and the lower portion of Konig’s curve approximates a simple relationship such that a tenfold in- crease in brightness produces a twofold in- crease in acuity. Recent data from the night training manual of the Royal Canadian Air Force show the relationship between acuity and the brightness of the sky against which planes are viewed. The RCAF figures, compiled by Evelyn, are more nearly comparable to the present situa- tion, and while only four points are shown the agreement between these data and Konig’s is good. Figure 1 shows that the curves from these two sources tend to be parallel, though they diverge somewhat at higher values. In the present study, corrections for illumi- nation were made by dividing the visible area of each target by the square of a value which varies continuously from 1 at 0.05 foot-lambert through 8 at 0.00005 foot-lambert, doubling with each tenfold decrease in brightness. The resulting quotient is the area of a hypothetical target illuminated by 0.05 foot-lambert and having a visibility equivalent to the one actually sighted. The difference in contrast for a given ship Table 1. Corrections of target length for angle of sighting. Angle between the lookout’s line of sight and the course of the target ship (degrees) Visible fraction of total length (per cent) 90-65 100 64-55 85 54-45 75 44-35 65 34-25 50 24-15 30 14- 0 20 ords are given in terms of equivalent square feet [ESF] or as the visual equivalent of a given area at the standard distance of 1,000 yards. Target Size. Even when the true dimensions of a target ship are known, the fact that all vessels have highly irregular profiles makes it impossible to do more than estimate the area of a visible silhouette. The type of ship (i.e., transport, cargo ship, tanker, or DE) was always known and average dimensions for these ships were available.5 The height used in calculating target area was measured from a point low enough on the ship so that masses large enough for spotting could be found at that level, since masts, fun- nels, and the like are rarely, if ever, of much use in picking up a target under night condi- tions. This practice necessitated the introduc- tion of an arbitrary figure for each type. Trans- ports were rated at height 50 feet, length 550 66 NIGHT LOOKOUTS between the situation in which it has sea for a background and that in which it has sky is sometimes very great. Consequently it was necessary to divide the total area of each ship into two parts: that part seen against a sea background and that seen against the sky. Determining the size of each of the two parts was a straightforward geometrical problem. The height of each lookout station was known, and the distance to the horizon from each height could be computed from the formula 1.15 V h = distance to horizon in nautical of the target appearing against sky and sea, the level of brightness, and the range of the target. For example, one of the after-sky-lookouts sighted a freighter at a relative bearing of 225 degrees. The freighter was 450 feet long, and its average visible height was 30 feet. But at the target angle at which the lookout spotted the vessel (45 degrees), only 75 per cent of its length was visible. The 450 feet was conse- quently reduced to 337.5 feet. The lookout’s station was 30 feet above the water and his horizon 12,770 yards away. At 5,000 yards (the range of the freighter) his line of sight to his horizon was 18 feet above the water. In other words, 18 feet of the target ship’s 30 feet of height were seen by the lookout sil- houetted against the sea, giving an area of 6,075 square feet (337.5 X 18 feet). The meas- ured brightness of the sea was 0.000028 foot- lambert, at which level acuity is only 1/9.3 as great as at the arbitrarily chosen standard illumination. The area of that portion of the ship viewed against the sea was, therefore, again adjusted as follows: 6,075/(9.3)2 = 70.2 square feet. The brightness of the sky, against which the other 12 feet of the target’s height were seen, was 0.0005 foot-lambert, and the divisor for acuity at that brightness is 8.0. Thus [337.5/(8.0)2] 12 feet or 63.4 square feet was the computed equivalent of the portion of the target ship seen above the horizon. The sum of these two terms gives the total visible area corrected for brightness: 70.2 -f- 63,4 = 134 square feet. It was still necessary to take account of the range at which the target was sighted and to reduce the area of 134 square feet to the area which would be visibly equiva- lent at a standard range of 1,000 yards. Thus 134 was divided by the quantity (5,000/1,000)2 giving 134/25 or 5,36 ESF. 6 2 2 Possible Sources of Error The five corrections illustrated above are subject to varying degrees of error. Some ships were undoubtedly longer or shorter, higher or lower, and more or less visible than they were rated. Some errors were inevitable in the meter readings, and the curve by which those read- Figure 1. Relation between level of illumination and visual acuity. miles, where h = lookout’s height above water in feet. The lookout’s line of sight to his hori- zon cuts the target ship at a height depending on the distance between the lookout and his target. That height could be calculated by simple trigonometric means and used for de- termining the area of sea-contrast. The area of sky-contrast equals the length of the ship times the remainder of its height. Summary of Treatment of Data Each ship reported by a lookout was reduced to the number of equivalent square feet [ESF] representing a hypothetical target (equivalent to the one actually seen) silhouetted against a sky of 0.05 foot-lambert brightness at a dis- tance of 1,000 yards. This reduction was ac- complished by correcting each raw report for size of ship, angle of appearance, proportions THE PERFORMANCE OF NIGHT LOOKOUTS ABOARD SHIP 67 ings were translated into equivalent visibilities was only an approximation to the true function relating brightness and acuity. In spite of these possibilities for error in the ESF unit, it was stable enough to give meaningful results in the present study, for the scores showed far greater ranges than could possibly be attributed to fluctuations in the measuring instrument. Another possibly significant variable was contrast. On clear moonlight nights the con- trast of the target would vary depending upon the position of the moon relative to the ob- server and the target. The influence of con- trast was greatly diminished in the present study by limiting the observation to cases where the target ship appeared as a dark object Figure 2. Distribution of single ESF measures. In addition to these defects three other fac- tors deserve comment. Haze is a factor of potential seriousness in this type of study. Unfortunately no method was available for measuring haze accurately. Its presence was recorded at the appropriate times but no way of quantifying its effect was found. Performance on hazy nights was later com- pared with performance on clear nights. By several methods of comparison it was shown that haze increased the minimum size a target must possess in order to be seen at a given distance under any specified light conditions. The amount of this effect is, of course, depend- ent upon the amount of haze. Since haze was not measured quantitatively, quantitative state- ments of its effect cannot be made. against a lighter background and by the fact that the angles of view of the lookouts prob- ably did not differ sufficiently to allow im- portant variations in contrast to appear. Under other operating conditions, however, differ- ences in contrast could be quite important. The possibility of errors due to the use of binoculars must also be considered. Such a pos- sibility results from two facts. First, although all lookouts had been instructed in methods of setting their glasses properly, it is difficult to reset glasses in the dark after another man has been using them, and the investigators had no way of checking on the adequacy of any man’s setting at the time of his test. Second, possible damage to binoculars was similarly beyond the control of the experimenters. In 68 NIGHT LOOKOUTS at least one case a pair was demonstrated to be faulty, but there was no way of discovering for how long lookouts’ reports from that station had been so handicapped. 623 Interpretation of the Data The data which were gathered and treated according to the procedures outlined above pro- vide tentative answers to three questions. The with extreme scores running up to very high values. The mean of these scores was 60 ESF. The range was from a best score of 3 ESF to a worst one of 650 ESF. When each man’s scores were averaged, the nature of the distribution changed only slightly. The extreme values dropped out as they were counterbalanced by more moderate scores, and the range narrowed to 5 to 316 ESF. The mean was practically unchanged, but the mode moved up to the region between 25 and 30 ESF, giving Figure 3. Distribution of average ESF measures. answers are tentative because of the limitations of the study. Nevertheless they are believed to be the best answers now available to these three questions: 1. What can the average lookout see at night ? 2. How much difference exists between a good night lookout and a poor one? 3. How consistent are lookouts? After records taken on certain nights had been discarded because of faulty operation of the low-brightness meter or for other technical reasons, there remained 308 usable records. These records were of reports made by 114 different men, some of whom had been tested only once, others as often as seven times. Fig- ure 2 gives the distribution of the 308 scores. The distribution was extremely skewed with a modal region between 10 and 20 ESF and the distribution a slightly more normal appear- ance. By and large, however, the distribution of single scores and the distribution of average scores do not differ greatly. Figure 3 displays the latter distribution. What Can the Average Lookout See at Night? Those concerned with the problem of select- ing and training lookouts have often been handicapped by a lack of standards by which to judge the performance of their men. On the basis of the present study, it now seems feasible to estimate what an average lookout can do in the way of spotting ships at night. The average score of the men studied was very stable. The means of the total group and of all the subgroups examined fell between 59 and 65 ESF. It seems reasonable to accept, for THE PERFORMANCE OF NIGHT LOOKOUTS ABOARD SHIP 69 use as a practical yardstick, a median figure of about 60 ESF as typical of the main body of lookouts on the ship studied, though it should be borne in mind that there was considerable variability about this norm. This score of 60 ESF means that the typical lookout, when stationed low enough for good contrast, can see a target of 60 square feet at 1.000 yards against a brightly moonlit sky. At the same distance in clear starlight, the same man requires a target to be almost 1,000 square feet. At 2,000 yards in starlight, he requires a target of about 3,800 square feet, or some- thing like a PC seen broadside. At 2,500 yards in starlight, he would have trouble seeing a small cargo ship. At the same distance on an overcast night with no moon, he would require a ship the size of a very large freighter. On a really dark night, his maximum useful range is not much over 3,000 yards for even the largest ship. All these figures are, of course, subject to modification by weather, the lookout’s condi- tion, etc., and all will depend somewhat on the height of the lookout. Men higher up have a better chance of seeing the bow wave or wake of fast moving ships, but the men lower down will get the greatest advantage from contrast against the sky; the typical lookout, whose effectiveness is exhausted at about 3,000 yards anyway, will only be handicapped on dark nights by being stationed high above the water in order to increase the distance of his horizon. How Much Difference Exists between a Good Night Lookout and a Poor One? The range was very great. The 308 indi- vidual measures scattered from 3 ESF to 784 ESF. The range of average measures for the 114 men was naturally somewhat smaller, but was still considerable. If only those men who were tested five or more times are considered, thus limiting the population to the fairly re- liable measures, the averages ranged from 210 ESF for the poorest man to 14 ESF for the best. This pair of figures means that the best man, using 7 X 50 binoculars, should be able to see a target of only 14 square feet at a distance of 1.000 yards in bright moonlight, while the worst one would require a target about 15 times as large, or 210 square feet in size, under the same conditions. In terms of distance, this difference means that the better of these two men can see a ship nearly four times as far away as can the poorer one. Table 2 gives some comparative areas for the best, average, and poorest lookouts under dif- ferent conditions of brightness and distance. For the purpose of translating these areas into ship sizes, the typical areas used in this study for different types of ship viewed broadside may be useful: large transports, 30,000 square feet; cargo ships, 12,000 to 20,000 square feet; tankers, 4,000 to 9,000 square feet; DE, 8,000 to 9,000 square feet. In practice, of course, certain types of ship are more easily visible than others of the same area. Table 2. The performance of lookouts.* Range in yards Illumination of sky, in ESF Overcast Clear starlight starlight Bright moonlight 22,400 1,400 10,000 6,000 21,000 22,400 5,600 350 5,000 24,000 1,500 5,250 3,584 896 56 2,000 15,360 3,840 240 53,750* 13,440 840 896 224 14 1,000 3,840 960 60 13,440 3,360 210 * All values are in equivalent square feet (see text). The first value in each cell is for the best lookout studied, the second for the average lookout, and the third for the poorest lookout. A line of dots in place of a figure indicates a value far too large for any existing ship. The asterisk indicates a value greater than any but the very largest passenger liners. A study of the comparisons offered in Table 2 should make it abundantly clear that there are very significant differences between the very poor night lookout and the expert. The differences obtained in this investigation, more- over, may not have been maximal, for there was reason to believe that even the best of the look- outs tested had not been as thoroughly or as carefully trained as possible. With differences as large as those reported here, the care used in selecting and training of lookouts could be multiplied many times before 70 NIGHT LOOKOUTS there would be any question of seriously dimin- ishing returns. If a good selection and training program could bring the entire lookout watch from the level of the average man reported here to that of the best, it would increase the range of the “eyes of the ship” to more than double its present range. Success in bringing the poor- est men up to the present average would have an even larger comparative effect. How Consistent Are Lookouts? Thirty of the subjects of this experiment were tested four or more times. With such a group, the comparison of two random samples of measures for the same men gives a fair esti- mate of the consistency with which the subjects performed. The correlation coefficient was cal- instance to take the logarithm of the ESF scores. The logarithm of each ESF score was taken and the average log ESF computed for each in- dividual, Figure 4 is the frequency polygon constructed from these data. The distribution is much more normal in appearance than the distribution of arithmetic scores shown in Fig- ure 3. It has a mean at 1.49 log ESF and a standard deviation of 0.40 log ESF. An even more striking change in the size of the reliability coefficient takes place when a logarithmic scale is used. The logarithm of each ESF measure was computed for the 30 men on whom four or more measures were obtained. The average logarithm of the odd-numbered measures was then correlated with the average N = 115 M = 1.49 LOG ESF 38-39-46 Training Aids Several training aids were developed32-34 and instructions were prepared for the most effec- tive use of training aids.19 Two types of drill material were devised. The first consisted of drill-board mock-ups of such units as the dis- tilling plant or the refrigeration system. These drill boards simulated both individual units and larger systems of equipment. They pro- vided an opportunity for drill on operating procedures where neither the shipboard equip- ment nor models of that equipment were available. The mock-ups were used in instruc- tion on theory as well as for drill in operating procedures. A further development was the operational drill chart. This type of training aid was ap- plicable to a large variety of engineering equipment systems. In consisted of a diagram- matic representation of engineering equipment on which operating controls were identified, and it illustrated what happened when those controls were manipulated. Operational drill charts were prepared for use in teaching the operation of the Soloshell distilling plant.32 Drill charts of this nature may be used (1) for familiarizing the trainee with operating procedures prior to actual drill; (2) for follow- ing the progress of a drill on a mock-up or on actual equipment; and (3) as a basis for dis- cussion of operating steps. -1 INSTRUCTOR TRAINING The course outlines, lesson plans, and train- ing aids listed above were all developed to assist the instructor in giving more effective instruction. In addition to these aids, a need was felt for more careful training of the in- structor himself. Courses on How to Teach Gunnery11 and How to Teach Engineering17 were developed for this purpose. Each of these courses was given to classes made up of Navy instructors. Each course was accompanied by a kit of training aids and by recorded samples of lectures, explanations, and drills which illustrated good and bad teaching technique. These samples were played to classes RESTRICTED 82 TRAINING NAVY GUNNERS AND ENGINE ROOM CREWS of instructors and used as a basis for their discussion and criticism. The two courses were much alike, but the illustrations and examples of principles were specific in each to the field of instruction for which it was intended. The course for gunnery instructors can illustrate both. It consisted of five sessions, each IV2 hours long. Lesson 1 gave the background and purpose of the course and developed 11 principles of good instruction. Lesson 2 gave an opportunity for rating in- struction recordings in terms of the principles of good instruction developed in the first lesson. These ratings were analyzed and discussed by the group. The remainder of the lesson was de- voted to the use of lesson plans, preparation of charts, models, and other equipment, and other problems of planning a class session. Lesson 3 was devoted entirely to a discussion of the problems centering around instruction itself. The use of instructional aids, the value and use of questions, use of summaries, the im- portance of job analysis, and how to conduct a demonstration were discussed. Lesson U was a discussion of the problems encountered in conducting practice and drill sessions. Lesson 5 was a review of the course. Re- cordings of instruction by members of the group which had been obtained well in advance of this lesson were played and analyzed. A check list outline of how to teach gunnery was used to point out good and poor features of instruction and to summarize and evaluate the course. It also furnished a method for the in- structor to use in auditing and evaluating his instruction at a later time. A small pamphlet, Is Your Teaching Effec- live,17 provided a basis for self-rating on the points emphasized in the two courses for in- structors. 75 CONCLUSION Throughout their work, project personnel sought to improve gunnery instruction and engineering instruction. The principal features of this work were: 1. Basing instruction on an analysis of the actual operating jobs. Job analysis of operat- ing procedures provided the basic information for writing any set of lesson plans. 2. Separating instruction levels so that sepa- rate and appropriate attention and emphasis could be given to (a) general information on the equipment to be used, (b) detailed study of specific operations, (c) drill on separate procedures, and (d) drill as part of an entire team or watch. 3. Using mock-ups, drill boards, models, trainers, and any other available device to help prepare the trainee for his actual operating duties. 4. Emphasizing operational drill rather than theory, 5. The lesson plans and course outlines were written to make effective use of motivation, repetition, drill, and other principles and tech- niques of good instruction. The training manuals were generally useful. Some were reproduced by the Bureau of Naval Personnel. Others were printed by the Opera- tional Training Commands, Eleven of them were distributed by the Navy to the fleet, to operational training and antiaircraft centers, and to all new ships at the time of commis- sioning. Chapter 8 TRAINING WINCH OPERATORS By Dael WolfleA SUMMARY A trainer consisting of a miniature electric winch was constructed to help train hatch- men and winchmen for duty on APA’s and AKA’s. 81 INTRODUCTION At the request of the Commander, Opera- tional Training Command, U. S. Pacific Fleet [COTCPac] a project was established to im- prove the selection and training of hatchmen and winchmen specialist teams assigned to duty on assault 'personnel auxiliary [APA] and assault cargo auxiliary [AKA] ships. Research written with the University of Southern Cali- fornia under NDRC Project N-116. Work began in July 1944 and continued through February 1945. By that time most of the AKA’s and APA’s were commissioned. 82 THE ELECTRIC WINCH TRAINER Project personnel designed an experimental model of an electric winch trainer4 in which the controls were of the same dimension as the ones found on electric winches being installed at that time on AKA’s and APA’s. The winches were otherwise miniature in size, speed, and lifting power. They were used in conjunction with miniature booms and rigged yard and stay. Figure 1. The electric winch trainer, controls. studies on hatchmen and winchmen were con- sidered urgent by COTCPac because of the large number of AKA’s and APA’s being built for duty in the Pacific and the small number of available men with civilian experience in han- dling winches. A contract for the work was The miniature electric winch trainer is shown in Figures 1 and 2. This equipment was tried out at the Small Craft Training Center Seamanship School on Roosevelt Base, Terminal Island, Cal. Follow- ing this trial use, a manual for instructors was developed.1 The manual contained a tested set of six lectures and practice drills arranged a This chapter is based on the work of NDRC Project N-116. 83 84 TRAINING WINCH OPERATORS according to the part-progressive method of instruction. The manual included a winch oper- ation test and an objective rating scale for the use of the instructor. It also described the con- struction and maintenance of the entire rig. Several attempts were made to conduct vali- trained men and details of untrained men who could be made available at the time an AKA or APA was available. A suggestive study was finally completed in which 13 trained men and 14 untrained men were compared in their first attempts to master Figure 2. The electric winch trainer in operation dation experiments. All but one of these en- countered scheduling difficulties. Validation of the trainer required testing the ability to handle standard winches on board ship of men who had had experience on the miniature trainer and of other men who had not had that experi- ence. Thus it was necessary to find details of the operation of standard electric winches aboard ship.2 The miniature-trained group was superior to the untrained group to a degree which was, despite the small number of cases, significant at the 1 per cent confidence level. The distributions of ratings3 given to trained and untrained groups are shown in Table 1. THE ELECTRIC WINCH TRAINER 85 Table 1. Distribution of proficiency ratings. 13 trained men 14 untrained men Proficiency rating (Number of (Number of ratings) ratings) 5.0 5 4.5 3 4 4.0 10 5 3.5 6 4 3.0 11 7 2.5 4 6 2.0 8 1.5 4 1.0 1 0.5 1 0.0 2 Median 3.9 2.9 Mean 3.7 2.7 In this table a comparison is given of the ability of 13 trained and 14 untrained men to handle a full-sized electric winch on an assault cargo auxiliary ship. The trained men had re- ceived their instruction on an electric winch trainer. Three judges made independent proficiency ratings of each man. The judges did not know until after the ratings were completed which of the 27 men had had prior training on the miniature electric winch trainer. Correlations of ratings by the different judges were 0.81, 0.83, and 0.89. Two additional trainers were built for in- stallation and use at COTCPac training sta- tions. Chapter 9 TRAINING AMPHIBIOUS CRAFT CREWS Dael Wolflea SUMMARY Assistance was given to the Amphibious f\_ Training Command in the development and improvement of training courses for the crews of amphibious craft. The work was of two types, development of specific courses of in- struction and collection of relevant information in order to improve instruction. Training courses were developed: 1. To instruct amphibious training base in- structors in effective teaching methods. 2. To train instructors in methods of measur- ing student achievement. 3. To instruct a specially selected group of classification specialists in the elements of per- sonality analysis, in order to improve the proc- essing of atypical cases such as psychiatric cases, illiterates, and billet misassignments. 4. To train attack boat personnel in gas de- fense and piloting. Two related programs provided information with which to improve instruction: 1. A questionnaire was administered to per- sonnel returning from combat zones in order to secure information about the effectiveness of various methods of training, classification, or ship performance in actual combat. 2. A job analysis was made of the amphibious enlisted billets. The training studies and the varied curricula were organized into a systematic overall pro- gram described in the Amphibious Training Command Training Manual. INTRODUCTION At the request of the Commander, Amphib- ious Training Command, U. S. Atlantic Fleet, NDRC Project N-117 was established on February 1, 1944, to aid the Command in the development of its personnel program. World War II necessitated the development of amphibious operations on a much larger scale than ever before; new ships, new tactics, new duties all produced new problems. The Amphib- ious Training Command was established to solve these problems and to train crews for the amphibious forces. The request to NDRC was for help in this undertaking. Speed was even more necessary than usual in this case. Preparations were being made and crews being trained for Atlantic and Pacific landings. The time schedule permitted very little research. Practically all the project’s efforts consisted of immediate application of ideas and techniques worked out at earlier times. The project was established at Amphibious Training Command headquarters but also worked at the amphibious training bases [ATB]. In addition to work on billet analysis and instruction described in this chapter, proj- ect personnel cooperated with Amphibious Training Command and Bureau of Naval Per- sonnel officers in developing standardized clas- sification procedures for officers and enlisted men (see the Summary Technical Report of the Applied Psychology Panel, Volume 1, Chap- ters 11 and 13). 92 TRAINING COURSES FOR AMPHIBIOUS TRAINING COMMAND Special training courses were developed in four fields, each of which is briefly described below. 9.2.x Training of Amphibious Training Base Instructors With the cooperation of NDRC Projects N-105 and NR-106 and the instructor training staff of the Bureau of Naval Personnel, a course on effective instruction was developed. The course was modeled after the course How to Teach Gunnery, described in Chapter 7, but a This chapter is based primarily upon the work of NDRC Project N-117. 86 TRAINING COURSES FOR AMPHIBIOUS TRAINING COMMAND 87 was specifically adapted to the problems en- countered by ATB instructors. The course con- sisted of ten lessons of from 1 to 2 hours in length. It was given to over 700 instructors of the Amphibious Training Command. The instructor training staff of the Bureau of Naval Personnel performed most of the work in this program with the administrative cooperation of Project N-117. After the course was developed and had been taught to several groups of instructors, the project withdrew. The program was carried on by the instructor training staff. This course was but one phase of a larger program for improvement of instruction. The whole program included voice recording of actual periods of instruction with subsequent analysis of the presentation by the instructor and a member of the instructor training staff, joint action with the base training aids officer to effect full utilization of existing training materials and to develop new aids, and periodic conferences with the base instructional staff. 9-2 2 Achievement Testing in the Amphibi- ous Training Bases In cooperation with the training staff at each base, an achievement testing program for offi- cers and men in training was initiated at ATB, Camp Bradford, Virginia, and at ATB, Little Creek, Virginia. A board of senior instructors from each of the training divisions was ap- pointed to supervise the construction and ad- ministration of achievement tests throughout the base. At the outset, a series of weekly con- ferences with this board was held by members of the project staff. The discussion covered spe- cific examination problems of the various train- ing divisions. The purpose of the conferences was to agree upon standard examination pro- cedures and to construct tests of known and appropriate difficulty which would adequately cover the subject matter. The participants con- structed specimen examinations in their subject fields which were analyzed and evaluated by the group. The manual, Measuring the Achieve- ment of the Trainee,2 served as the basis for much of this conference work. The base training- officer and the board of senior instructors continued to supervise ex- amination procedures at each base after project personnel withdrew from this activity. The adoption of standardized achievement tests increased the motivation of instructors and trainees and improved measurement of knowledge of course content and operational achievement.2 At the request of the Standards and Curricu- lum Division of the Bureau of Naval Personnel, a representative of Project N-106 assisted in the development of an achievement testing pro- gram at ATB, Coronado, California.10 Fourteen minimum essentials achievement tests were prepared in order to evaluate the success of the training program. Eleven of the tests were for use in the ship-to-shore training, two were gunnery tests, and one was for use in the boat training division. Reliabilities of seven of the tests were computed. The coefficients varied from .74 to .94. Distributions of scores were made for ex- aminations administered after 4 weeks’ train- ing, The distributions indicated that practically all students failed to attain the minimal essen- tials prescribed by the curriculum for blinker receiving and compass-and-steering. Achieve- ment was best on semaphore sending, maneu- vering signal (semaphore), and beach marker tests. In general, the achievement of students was poorer than had been anticipated in view of the ease of the material taught and the average ability of the students. The difficulty of each item in the 14 tests was determined by finding the percentage of students who passed it. The results indicated specific areas where instruction was poor. Sug- gestions were made for the improvement of teaching in some of these areas. Course in Personality Analysis Early in the classification of enlisted per- sonnel at the ATB’s, a need was felt for special- ized processing of atypical cases, such as illiter- ates, billet misassignments, and men suspected of being psychiatric cases. Accordingly, Project N-117 in cooperation with the force surgeon 88 TRAINING AMPHIBIOUS CRAFT CREWS developed and administered a course of instruc- tion for selected specialists (C), intended to provide each classification office with at least one classification interviewer particularly trained to screen out such cases for further consideration.7 Nine specialists (C), drawn from the four Atlantic Coast training bases, were given a course in personality analysis at the ATB, Camp Bradford, Virginia. The course included elementary instruction in the nature of psychology, individual differences, emotions, personality, the measurement of intelligence and other traits, Navy tests, and psychiatry. Instruction in psychiatry was given by the base medical officer and the staff of the Portsmouth U. S. Naval Hospital; it included clinical obser- vation of maladjusted and combat fatigue cases. Instruction in the other topics was given by project personnel. Upon completing the course, the specialists (C) were designated as special case technicians by the Amphibious Training Command and were assigned to the mobile classification unit and to the ATB classification offices. They were provided with interviewing rooms, reference libraries, and special psychological testing equipment to aid them in screening atypical cases for referral to the base medical officers for further consideration. Accurate evaluation of the work of the special case technicians would have necessitated a study in which the base medical officer would have diagnosed both those cases which the tech- nicians suspected of having emotional malad- justments and so referred to him for interview and those processed by the technicians but not so referred. Such a study would have made it possible to determine what percentage of re- ferred cases was correctly referred and what percentage of nonreferred cases should have been referred. Owing to the heavy case loads of the base medical staffs, the study was not pos- sible. However, for two of the training bases and for the mobile classification unit, determi- nations were made of the percentages of psy- chiatric referrals which the base medical officer also diagnosed as emotionally disqualified. The figures are presented below. At two establishments over 60 per cent of men referred to the base medical officers were diagnosed as emotionally disqualified. At the third establishment 36 per cent were disquali- fied, These results indicated that the special case technicians did a reasonably accurate job of screening possible psychiatric cases for refer- ral. The special training given them seemed justified. Enlisted men processed by special case technicians Psychiatric referrals by special case technicians Psychiatric referrals disqualified by medical officers ATB, Camp Bradford 5,443 570 354 (62.1%) ATB, Little Creek 5,119 1,400 512 (36.6%) Mobile classi- fication unit 1,574 1,259 816 (64.8%) At the request of the Commander, Training Command, Amphibious Force, U. S. Pacific Fleet, the course in Personality Analysis for Specialists (C), Special Case Technicians was administered by the project to a group of five specialists (C) at ATB, Coronado, California, during November 1944. 9'2A Courses in Gas Defense and Piloting The project staff cooperated with the train- ing officer, ATB, Fort Pierce, Florida, in the construction of courses in gas defense and piloting specifically designed for the training of attack boat personnel.5 The courses featured active participation by the trainees, largely through the use of drills and problem work sheets. 93 INFORMATION TO IMPROVE INSTRUCTION Two types of work were undertaken in order to secure information for improving the classi- fication and training programs of the Amphibi- ous Training Command. 1. A questionnaire was administered to per- sonnel returning from combat zones. 2. A job analysis was made of the billets on amphibious craft. Each is briefly described below. THE AMPHIBIOUS TRAINING COMMAND TRAINING MANUAL 89 931 Questionnaire for Personnel Return- ing from Combat Zones As one means of obtaining information which might prove of value in the amphibious train- ing program, the project, in cooperation with the Amphibious Training Command, con- structed and administered a questionnaire of 93 items for personnel returning from combat zones.4 The project analyzed the replies made by 511 enlisted men and 141 officers returning from amphibious duty and by 164 enlisted men returning from nonamphibious duty. Tabula- tion of the responses gave information regard- ing the men’s experience and opinions in respect to training, classification, ship perform- ance, morale, gear, supplies, etc. Most of the officers and an even larger frac- tion of the enlisted men considered their train- ing satisfactory for the duties they were as- signed aboard ship. Some dissatisfaction was revealed, however, over the way in which train- ing time was used. For example, half the officers and about 40 per cent of the men reported in- adequate instruction in the handling of perish- able supplies. The tabulated results, the individual ques- tionnaire forms, and specific suggestions for improvements which were made by many of the men were furnished to administrative offi- cers of the Amphibious Training Command. These data were useful in indicating ways in which amphibious training could be made more satisfactory. 9 3 2 Analysis of Amphibious Enlisted Billets The project, at the request of the Amphibious Training Command, undertook to analyze the duties of amphibious enlisted billets on four types of ships :6 landing ship tank [LST] ; landing ship medium [LSM] ; landing craft infantry [LCI] ; and attack boats. The Com- mand assigned four officers to work with the project in making the billet analyses. The analysis form consisted of a summary of billet duties and an activity analysis. The analyses indicated the functions which were considered necessary to insure successful per- formance of billet duties but did not provide detailed instructions for carrying out those duties. The analyses were used in the develop- ment and revision of courses and in some cases of the watch, quarter, and station bills. They proved particularly useful in planning instruc- tion whenever a base had to start training per- sonnel for ships or, craft which were new to that base. 94 THE AMPHIBIOUS TRAINING COMMAND TRAINING MANUAL The training studies and the varied curricula of the Amphibious Training Command, U. S. Atlantic Fleet, were organized into a systematic overall program described in the Amphibious Training Command Training Manual.3 The training manual consisted of two parts. Part II outlined the curricula themselves in order to effect standardized training in the several ATB’s. Part I of the training manual was prepared cooperatively by Amphibious Training Command officers and project per- sonnel. It provided a statement of the objectives of the training program and the means of achieving those objectives. It complemented the Manual of Classification Procedures for Am- phibious Training Basest also cooperatively developed by Amphibious Training Command and project personnel, and utilized the results of the project’s extensive study, The Effective- ness of Classification Data in Predicting Billet Performance in Training in the Amphibious Force? It incorporated the studies reported in this chapter into the organization and adminis- tration of the whole training program. Chapter 10 TRAINING NAVY TELEPHONE TALKERS By Dael Wolfie11 SUMMARY The high noise levels encountered aboard ship frequently made normal speech unin- telligible over shipboard telephone circuits. Special training courses were developed in order to increase the intelligibility of telephone com- munications. A telephone talkers’ manual, a manual for instructors, and phonograph records were de- veloped for use in a course for telephone talkers. Experimental investigations of the best meth- ods of instruction and of the value of various parts of the course and a survey of the training given to telephone talkers in a number of in- stallations led to improved standardized in- struction. The training ashore was integrated with subsequent training on board ship. Three courses were developed for the sub- marine service: a basic course to teach general skills; an intermediate course to give each man proficiency in the use of the phraseology re- quired by his assignment aboard ship; and an advanced course to drill the entire crew as a combat team in the coordinated use of the communication circuits. 101 INTRODUCTION Speaking over a telephone is a relatively simple and commonplace activity. Yet special training in the use of shipboard telephone equip- ment was necessary. There were two reasons. One was the fact that shipboard phones were sound-powered, depending entirely upon the en- ergy of the speaker’s voice to get a message through. The second reason was the high noise level in which talking was necessary. The differ- ent equipment and the difficult conditions both meant that normal telephone speaking habits had to be overcome. Hence special training courses were a necessity. At higher noise levels, the intelligibility of speech over sound-powered phones drops very sharply. This drop in intelligibility was clearly shown in an experiment9 in which trained talk- ers read lists of familiar words over sound- powered phones to trained listeners. A noise approximately the same as diesel noise in qual- ity was used as interference. The results are shown below. At a noise level of 110 decibels, such as exists in a submarine engine room with both diesels running, 73 per cent of the words were missed. These results, obtained with expert talkers and expert listeners, make obvious the difficulty of passing messages on board ship under comparable circumstances. Noise Words correct Words missed (db) (%) (%) None 91 9 90 77 23 100 60 40 110 27 73 120 13 87 Recognizing the difficulties involved in voice communication aboard ship, the Navy requested NDRC help in solving those problems. The re- quest was accepted and assigned as Project N-109 to the Psychological Corporation for study. Work on the selection and training of shipboard telephone talkers was started early in 1943. Chronologically, the work was divided into three fairly separate phases. The first phase was largely exploratory and covered both selection and training.1’3> 4 The speech interview, developed to aid in selecting Navy telephone talkers, is described in the Summary Technical Report of the Applied Psychology Panel, Volume 1, Chapter 10. The initial work on training resulted in the devel- opment of a short course of instruction for tele- phone talkers and several manuals and training aids. These are described in Section 10.2 of this chapter. The second phase consisted of extending and adapting for submarine use the techniques al- ready developed for training talkers on surface ships. This work is described in Section 10.5. The final phase (with surface ships) con- a This chapter is based upon the work of NDRC Project N-109. 90 TRAINING TELEPHONE TALKERS 91 sisted primarily of improving and standardiz- ing the many telephone talker courses which had grown up largely as a result of the work done in the first phase. Because of the logical relation to the first work, these developments are discussed, out of chronological order, in Sections 10.2.3, 10.3, and 10.4. In general, the manner of speaking and the way of using communication equipment found most satisfactory on shipboard were the same as those found most satisfactory in airplanes. Since these procedures are discussed in detail in Chapter 23, they will not be considered here. 102 TRAINING TELEPHONE TALKERS Experimental Comparison of Methods of Instruction Preliminary trials1 indicated that short pe- riods of training sometimes produced marked improvements in intelligibility over sound- powered phones. A more comprehensive experi- ment was therefore conducted to determine, if possible, the relative value of various methods for training recruits to speak intelligibly over sound-powered telephone circuits.2 Fourteen methods of instruction were tried out. They were of four general types: (1) in- struction through materials which the recruit read silently; (2) instruction by means of phonograph recordings; (3) instruction by teachers over phones; and (4) instruction in a teacher-classroom or face-to-face situation. In addition, control groups were used to check the relative effectiveness of the training methods. The procedure followed was to select 17 re- cruits as a listening panel and to divide the remaining recruits into two groups of speakers. The 17 men in the panel were given instruction over the sound-powered phones, all phones being in parallel so that the entire panel could be instructed at the same time. After hearing these instructions, the listening panel was divided into two groups, one panel listening to one group of speakers, the other to the second group. Each speaker read from one of a series of nine cards. Each card contained 36 randomly selected digits, arranged in 12 series of three. Alternate series followed the words “bearing” and “range.” These words were used merely to give the listeners time to write the digits, for example, “Bearing 6 3 1, range 9 4 2.” The listening panels recorded the digits as they heard them, using a mimeographed blank prepared for this purpose. Both listening panel members and speakers were recruits at the Naval Training Station, Bainbridge, Maryland. Experiments were normally divided into three parts: (1) an intelligibility test, (2) in- struction, and (3) another intelligibility test. In the pretest, instruction for the speaker was kept at a minimum. He was shown how to put on the earphones, told to talk with his mouth 1/2 inch from the transmitter, and instructed to read the words and numbers in a loud, clear voice. The only coaching given was to move the trans- mitter closer to the speaker’s mouth and to caution him to read more slowly in case such directions were needed. During the second, or instruction, period dif- ferent training methods were introduced for different groups of subjects. The retest period involved a repetition of the pretest procedure, the purpose being to deter- mine the effectiveness of the instruction in in- creasing intelligibility. The increases in intelligibility resulting from the different training methods were compared individually by the t-statistic to determine the relative reliability of the differences found. A control series, consisting of a pretest and retest with no intervening training, was used as a standard for evaluating all methods. All methods of instruction, including the con- trol series, resulted in definite improvement in intelligibility. The instructional method which produced the greatest improvement was one called “mass drill.” It was therefore adopted as the basic classroom procedure for the telephone talker course. The brief directions given the in- structor, his introduction to a new class, and the start of the drill itself are quoted below. Note to the Instructor The following drill is to be read to the men in a lively spirited fashion, looking at them as much as possible. Soon you will have memorized the drill, which should give you a better contact with your audience than you had while reading the drill. 92 TRAINING NAVY TELEPHONE TALKERS The class must participate. If they don’t follow you at first, have them try the drill sentence again until you are satisfied. This drill has been used with a large number of re- cruits. Properly administered it becomes a successful teaching device. Mass Drill Every ship has at least two phone circuits. Large ships have as many as a small city. These phone circuits connect every part of the ship. On each circuit are a number of telephone talkers, whose important job is to pass orders and information to different stations. The outcome of a sea battle depends in part on the ability of these talkers. Poor talkers who make “repeats” neces- sary may sink a ship. Good talkers may save a ship and help in sinking enemy craft. Sound-powered phones are used in the Navy. The good talker on a sound-powered phone must be alert in listening to and repeating commands. He must be able to remember orders so that he can repeat them cor- rectly. And very important, he must speak clearly, as time during action at sea is too valuable to allow for “repeats.” Now the question is, What makes a good telephone talker? First of all, the good speaker is loud, because a sound- powered phone gets its power from the voice alone. If you talk over the sound-powered phone as you might to a friend standing next to you, your voice might not be heard at all. “The louder the better” is a good rule in this case. Let us see for the moment how loudly some of you speak. All of you say after me: “One, two, thu-ree, four, fi-ive, six, seven, eight, niner.” Now say, only much louder, “Range four . . . fi-ive . . . oh , . . double . . . oh.” (Pick out various individuals to say this.) Don’t worry about talking too loud! Very few men talk too loudly over sound-powered phones. There is little or none of the blasting or fuzzing effect you hear on battery phones. In the second place, loudness control is important in good telephone talkers. By loudness control, we mean that the loudness of the voice is continuous, rather than trailing off at the end of a phrase or sentence. If the listener cannot hear the last word of a message, the whole message may have to be repeated! Do not speak this way, with the last part of the mes- sage almost unintelligible, “Control from sky lookout, plane sighted off starboard bow.” (Let voice trail off.) Instead, try to say it this way, with sustained loud- ness, “Control from sky lookout, plane sighted off starboard bow,” Now all together. Now you over there. (Point out a few men and have them repeat after you.) io.2.2 draining Manuals and Training Aids Telephone Talkers’ Manual Project personnel cooperated with officers from a number of Navy offices interested in communication problems (COMINCH, COTCL- ant, BuPers, and the Interior Control Board) in the production of a telephone talkers’ manual for the U. S. Fleet.11 The manual was put in final form in cooperation with the Training Aids Division, BuPers, and was issued as a publication of the Headquarters of the Com- mander-in-Chief, U. S. Fleet, dated November 19, 1943. The manual was prepared in quantity by the Bureau of Naval Personnel and was distributed widely to the fleet and to shore stations. A re- vised edition embodying some minor changes but keeping the general content and format of the original manual was prepared and distrib- uted later by the Bureau of Naval Personnel. Instructors’ Guide The U. 3 Instructors’ Handbook for a Course in Voice Communication for Pilot Instructors (U Hours).12 Indoctrination in Voice Communication for Instructors in Primary Flying Schools.1* Instructors’ Syllabi for Courses in Interphone Communication for Navigators and Bombard- iers.11 Instructors’ Syllabus for a Radio Operator- Mechanic Course in Voice Communication.c Students’ Workbook, Radio Communications, Basic.2* Instructors’ Handbook: Basic Radio Com- munications24 In the speaking exercises students on each party line worked as a unit. In most cases the sequence of messages in an exercise made an entity, for example the radio and/or interphone messages of a simulated flight. As one person spoke, the others on the line listened. Frequently the message designated which listener should respond. Each exercise had a dual purpose, to establish habits for intelligibility and to estab- lish habitual use of routinized message forms. The following excerpt from a drill shows how some of the materials were given to the instruc- tor. He, in turn, reproduced them in usable form for the individual stations. Drill 25: R/T Position Reports This drill is a simulated airway flight. Plane No. 3459 is cruising at 200 miles per hour. Each position report should include, in order, call, position relative to check point, time, altitude, flight conditions, ETA next check point or destination. Example: Station 1: Memphis Radio, this is Army 3459, over. Station 7: Army 3459, this is Memphis Radio, over. Station 1: Memphis Radio, this is Army 3459, 15 miles northeast Memphis, time 0935 at 4000 on instruments, estimate Texarkana time 1055, over. Station 7: Army 3459, Roger, out. Station acting Station acting as as Range Additional plane range station En route information 1 7 Memphis Nash- ville to Dallas 260 miles to Tex- arkana 2 8 Texarkana 100 miles to Dallas 3 9 Dallas Request permission to contact Love Tower During the last half-hour of the course, as during the first, an intelligibility test was given. SPEECH TRAINING COURSES FOR THE AAF 103 Manuals for Instruction in Voice Communica- tion for Aerial Gunnery.4 Students’ Manual for a Basic Course in Inter- phone Communication in Flexible Gunnery Schools.22 Instructors’ Handbook for a Course in Voice Communication for Control Tower Operators (8 Hours) M In addition to these, special drills in voice communication procedure were written for spe- cial purposes, such as bombing through over- cast,26 night fighter operations,16 and the Far East Air Forces16; and sections were written for various aircrew information files, such as pilots information file and radio operators in- formation file. One of the final tasks of the Voice Communi- cation Project was to revise these training manuals and instructors’ handbooks and to bring both methods and contents up to date (July 1945). The various manuals were com- bined into a single handbook for instructors and a set of specialized drills for different mem- bers of the aircrew.16 This task was completed a few days before World War II ended; the new combined manual was therefore not printed. The manuscript copy was given to the Assistant Chief of Air Staff-3 for use by the AAF Training Command. This new manual, rather than the various earlier ones named above, is reproduced on microfilm to accompany the Applied Psychology Panel’s Summary Tech- nical Report. 112-4 Crew Training In order to provide materials for training the crews of large bombers in interphone pro- cedure, the project developed an interphone crew trainer17 and special drills for crew train- ing.16 Each Crew Commander25 conducted the training course for his own crew. In most re- spects this course was very similar to those described in Section 11.2.2. 1125 Effects of Voice Communication Training The measured effect of voice communication training was increased intelligibility. There were other effects—better handling of equip- ment and use of standardized message forms— but their extent was not measured. Tests given at the beginning and end of each course regu- larly demonstrated increased intelligibility. Effect of Speech Drill The increase in intelligibility was in all prob- ability due more to improvements in speaking ability than to improvements in listening abil- ity.8- 20 Furthermore, the improvement resulted from the training course itself and not from unsuper- vised practice in the use of communication equipment. The effects of experience alone in changing speaking ability are shown in Table 1. Groups of instructors whose jobs necessitated teaching in the air, combat returnees, students in training, and men awaiting training (all stationed at one center) were tested for in- telligibility under the same circumstances. Their experience in flight and, consequently, in using communication equipment ranged from none to the large amounts represented by long periods instructing in an airplane or by completion of a tour of combat duty followed by reassignment to this country. In spite of these differences in experience, no significant differences in in- telligibility were found among the groups. The cadets in training were on the average as in- telligible as the men who had returned from combat. Table 1. Intelligibility scores for instructors, combat returnees, cadets, and preflight student navigators. Mean Group N intelligibility score* am Instructors 27 67.9 2.1 Combat returnees 20 60.2 2.6 Cadets 78 58.7 1.8 Preflight students 38 60.3 1.9 Total 163 60.7 * The noise level in the classroom in which these scores were oh- tained was somewhat below the recommended 100 to 110 decibel level. The mean intelligibility scores were therefore all substantially above 50. This fact does not invalidate comparisons among the groups. A survey of the frequency with which inter- phone messages had to be repeated in order to be understood further supports the idea that experience alone did not provide adequate voice communication skill. Among aircrews who had 104 VOICE COMMUNICATION TRAINING not had voice communication drill, but who were in other respects fully trained and ready to be sent to combat theaters, it was found that 40 per cent of interphone messages were re- peated one or more times. Many more were never acknowledged as received. tests were given to two classes 30 days after they were trained in voice communication. In each class a small but statistically significant improvement in mean intelligibility scores was found. The improvement may have been due to experience with the test itself, but it is clear that the improvement which results from training was retained without loss for at least a month. Effect of Instructor Figure 3 typifies the results of laboratory training conducted by speech teachers. If the program was to be of value it had to operate under the direction of Army instructors. Table 2 shows that the results obtained in field use by teachers who had little speech training other than 1 week’s indoctrination given by project personnel were about equally satisfactory. Al- though only five centers were surveyed thor- oughly, they appeared to represent what was happening generally in basic training installa- tions. Table 2. Intelligibility scores of speakers trained in voice communication at five AAF centers. Type of training center Untrained N mean Trained mean Mean gain Pilot 329 49.3 67.1 17.8 Pilot 128 59.6 76.7 17.1 Gunner 543 67.7 81.7 14.0 Navigator 229 65.9 85.4 19.5 Bombardier 78 68.6 84.7 16.1 Bringing All Men to Approximately Equal Intelligibility The effectiveness of the training program with regard to intelligibility is illustrated in an analysis of a representative class of 141 stu- dent pilots. Before training, the class had a mean intelligibility score of 52.6 (a, 11.4). After four hours of training the mean score was 70.0 (o, 8.5). Table 3 shows the gain of the class by deciles, divided according to pre-training intelligibility scores, and it also illustrates the decrease in variability that accompanied train- ing. This training made the men more intel- ligible and made them more alike in intelligi- bility. Figure 3. Effect of 4 hours of speech training on intelligibility over airplane communication equip- ment. Specific training in voice communication was necessary to improve message intelligibility. Retention of Increased Intelligibility Another aspect of the effectiveness of train- ing is retention of the acquired skill. Follow-up CONCLUSION 105 Table 3. ing upon ity. Effects of speakers of voice communication train- different pre-training abil- Decile Initial score Final score Gain 1 70.21 74.93 4.72 2 63.64 72.71 9.07 3 60.57 71.00 10.43 4 57.21 72.14 14.93 5 54.86 68.57 13.71 6 51.71 73.00 21.29 7 48.50 69.57 21.07 8 43.64 69.86 26.32 9 39.93 64.71 24.78 10 30.07 65.86 35.79 Total 52.6 70.0 17.4 of Plans and Operations Division, written to the Chief Signal Officer, General Meade wrote, “The Voice Communication Laboratory, lo- cated at Waco Army Air Field, and operated under NDRC Project SC-67, has developed training methods by means of which intelligibil- ity over the inter-phone and radio telephone may be increased on the average by as much as 25 per cent. ... It should be indicated that an average increase of 25 per cent in intelligibil- ity is greater than the increase that has been obtained in recent months through costly changes in equipment” (SPSOO 334, June 7, 1944, 3rd Ind., June 26, 1944). After the results reported in this chapter were made known, AAF directives required all aircrew personnel undergoing training in the United States to take a voice communication course. 113 CONCLUSION The importance attached by the AAF to the project’s success in improving voice communi- cation training was demonstrated by a letter from Brig. Gen. F. C. Meade, USA, Director Chapter 12 TRAINING RADIO OPERATORS By Dael Wolfie‘l SUMMARY Improvements were made in all phases of training radio code operators: initial learn- ing of the code characters; acquiring receiving speed; learning to send; measuring both re- ceiving and sending speed and skill; and the standardization of code speeds. The code-voice method of teaching basic code was developed on the basis of established prin- ciples of learning. It was shown to be superior to former methods and was adopted by the Army. During the weeks when receiving speed is being acquired, 4 hours of daily drill were shown to produce as rapid learning as 7. The distribution of these 4 hours within a day was found to be unimportant. Increasing the variety of drill materials was shown to produce more rapid learning and to decrease boredom in both students and instructors. One hour a day of practice in copying hand-sent, dear-text prac- tice material produced a small improvement in ability to copy cipher accurately. Giving men practice in copying code through various types of interfering noises did not diminish their ability to copy clear code and led to a moderate improvement in ability to copy code through interference. Work with high-speed operators consisted of an evaluation of two devices which had been suggested as ways of helping men to attain speeds of 25 words per minute. Both were shown to be without value. The introduction of sending practice early in the course was shown to have no detrimental effect upon the speed with which men learned to receive code. Since the earlier introduction allowed more practice in sending, this schedule was recommended. A trainer to aid men in learning to send correctly was developed. It consisted of a type- writer which was controlled by electronic cir- cuits in such a way that Morse code characters were transcribed by the typewriter as ordinary letters. Correct sending appeared immediately on the typewriter as correct copy. Errors in sending appeared as errors and usually in- formed the student immediately of the nature of the mistake he had made. A monograph on code speeds was prepared to provide instructors and others with an under- standing of the various bases for computing code speed and with instructions for cutting tapes which would have the exact speed de- sired. Two studies were made of errors, one of errors in receiving and one of errors in send- ing. In both cases the order of difficulty of the characters was found to be highly constant for students of different levels of ability and ad- vancement. Practice material, both for receiv- ing and for sending, in which difficult characters appeared more frequently than easy ones was suggested as a plausible method of improving learning. Neither suggestion had been ade- quately tested at the time NDRC’s work on code learning ended. The progress of several hundred students learning to receive code was recorded and tabulated. Tables and curves based on these data provided information on average rate of learning and upon variability of rate in learn- ing for students ranging in level from beginners to men able to receive at 25 groups per minute. Standardized tests of ability to receive code at different speeds were constructed. They were used in research studies and in Navy schools. 121 INTRODUCTION One of the very first requests made of the Committee on Service Personnel was to assist the Navy in improving methods of selecting and training radio code operators. The request was answered by establishing Project N-107, Later, two other radio code projects were activated: Project SC-88 worked on methods of training a This chapter is based primarily upon the work of NDRC Projects N-107, SC-88, and NS-366. 106 TRAINING MEN TO RECEIVE INTERNATIONAL xMORSE CODE 107 Army operators; Project NS-366 developed a Morse-code-actuated typewriter for use in im- proving sending skill. Code learning is an activity of considerable psychological interest for it provides an oppor- tunity to study a practically important kind of learning which is rather easily subjected to experimental study. There was, in consequence, a fair background of civilian research for these projects to draw upon. The first assignment for Project N-107 was to get acquainted with procedures being followed in 1942 in selecting, training, dismissing, and rating radio operator students. A survey was made of the practices followed in some twenty Army and Navy schools.1 Great diversity was found in organization, methods, and standards. For example, the 20 schools followed 11 differ- ent systems in introducing the individual char- acters and lesson groups. In one school all characters were introduced in the first 2 or 3 days; in another the men were in school for 6 weeks before encountering all 36 characters. Proficiency tests were given as frequently as once or twice a day and as infrequently as once a week. Despite a large amount of past experi- ence, there were no generally accepted require- ments, methods, or standards for training code operators. The radio code projects of NDRC undertook to determine the best methods in order that they might become standard prac- tice. Work on selection of radio operators is de- scribed in Chapter 6 of Volume 1 of the Sum- mary Technical Report of the Applied Psychol- ogy Panel. It culminated in the adoption by both Army and Navy of the NDRC-developed Speed of Response Test of Code Aptitude. Work on training is described in this chap- ter. It culminated in the adoption by the Army of a new manual25 on the training of code op- erators, a manual putting into standard use improved methods developed by project per- sonnel. 122 TRAINING MEN TO RECEIVE INTERNATIONAL MORSE CODE Code training consists of two phases. In the first, recognition of the characters representing the 36 letters and numbers must be learned. In the second, speed and skill must be acquired in recognizing the characters and in sending them. Learning the Characters The Code-Voice Method The first task in learning code is to learn to identify each of the 36 sound patterns (dots and dashes) which represent the 36 letters and digits. Various methods have been used to produce this learning. All consist essentially of helping the student to associate the sound of each character with its more familiar Eng- lish equivalent. Thus when the pattern • - is sounded, the student may be told that it repre- sents A (or Able), he may search through a printed list until he finds the correct letter, or he may have to wait until he has heard a num- ber of such patterns before he is informed of the equivalent of each. On the basis of general learning principles, it appeared likely that learning would proceed most rapidly if the learner was told immediately after hearing each character just what that character was. In order to provide him with an opportunity to recognize the characters for himself, a pause of about three seconds was introduced between sending the character and naming it. In order to allow a fairly large num- ber of correct responses, the characters were frequently sent in “doubles,” for example: • -, pause, “Able,” pause, • -, pause, “Able.” Then another character was sent, named, repeated, etc. This was called the code-voice method of teaching code.15’23'25 Again on the basis of general learning prin- ciples, it appeared likely that learning would proceed more rapidly if all 36 characters were learned than if they were broken up into groups, as had previously been common practice. “Whole learning” of all 36 characters at once was therefore included as a feature of the code- voice method. In order to provide the student with a record of his progress and at the same time to give him a standard method of scoring his own work, special record sheets were prepared. Each line 108 TRAINING RADIO OPERATORS consisted of a double row of squares, like this: the Army General Classification Test and in code ability as measured by the Army Radio Operator Aptitude Test [ROA]. The mean number of hours for 74 Z-tape stu- dents to qualify at 5 wpm was 35. The corre- sponding value for 253 code-voice students was 27. The percentage of Z-tape students who qualified at 5 wpm by noon of the seventh day of instruction was 29; for code-voice students, the mean was 50. Both differences were highly significant statistically, and in each case the superiority of the code-voice method was demon- strated, The data are summarized in Table 1. This demonstration led to the immediate adop- tion of the code-voice method at the Central Signal Corps School, Camp Crowder, Missouri, where the experiment was conducted. When a character was sent, the student wrote it, or wrote what he thought it to be, in the upper row, for example R in the first box. When the instructor announced the character, the student learned he was correct, wrote noth- ing in the lower row, and could respond ac- curately when R was repeated. When an error was made, as in the third box, or the student did not recognize the character at all, as in the fifth, the instructor’s naming it allowed im- mediate correction. When the missed character was repeated, a correct identification was al- most always possible. Counting the entries in the lower row of squares quickly gave the total number of errors. Plotting these error scores gave an easily made and easily understood learning curve. The code-voice method combined the psycho- logical advantages of immediate knowledge of results, whole learning, meaningful scores, a running record of improvement, and the oppor- tunity to compete with one’s own past record. Experimental Tests of the Code-Voice Method Inexperienced students of three beginning code classes were trained with standard Z-tape procedure, the usual Army method. Progress in code receiving was measured in terms of the number of hours required to learn to receive 5 words per minute [wpm] correctly and in terms of the percentage of men who qualified at 5 wpm or above by the end of the seventh day of training. Inexperienced students in the succeeding three entering classes were given code-voice training for 6 days and moved to the 5 wpm tape on the seventh day. No student, however proficient at the code-voice level, was permitted to practice on tape until the morning of the seventh day. Progress was measured by the same criteria used with the preceding three classes. There were 87 men in the Z-tape group and 262 in the code-voice group. The two groups were equal in general ability as measured by Table 1, methods . Comparison of code-voice and Z-tape of teaching International Morse Code. Training method Tape standard Number of men Hours required to pass 5 wpm Mean a Per cent who failed to pass 5 wpm Code-voice Paris 253 26.57 7.93 3.4 Code-voice Codez 258 25.66 14.59 Z-tape Paris 74 34.78 15.47 15.0 Z-tape Paris 446 40.87 22.78 Code-Voice Training Manual Following the adoption of the code-voice method at Camp Crowder, project personnel cooperated with the radio training section in preparing an instructors’ manual for the method-’3; an introductory pamphlet and ex- planation for the student24; and a WD manual, TM 11-459,25 which gave the code-voice method formal Army approval and made it a standard training method. Variants of the Code-Voice Method Situations occurred in which code learning could not be taught in regular classes or in which limited practice time, such as an hour a day, was available, A simplified code-voice pro- cedure was worked out for situations of this kind. An experimental test was made of the speed of learning of 77 soldiers taught 1 hour a day by this simplified method.15 In terms of TRAINING MEN TO RECEIVE INTERNATIONAL MORSE CODE 109 actual hours of practice, they learned slightly more rapidly than a normal school group taught by the standard code-voice procedure. The Office of the Chief Signal Officer and the Infantry School, Fort Benning, Georgia, pro- duced a Basic Radio Code course for the Armed Forces Institute. This course used the code- voice technique as its basic method. The course was recorded on phonograph records, making its use possible by students working without an instructor. An experimental evaluation of the Basic Radio Code course was attempted. Various dif- ficulties arose to interfere with the experiment, but the results indicated that the method was satisfactory and that the course could be recom- mended for use.0 12.2.2 Intermediate Stages of Learning Following the learning of the individual char- acters making up the International Morse Code, the would-be operator must learn to identify the characters as they are sent to him at faster and faster rates. Acquiring this skill takes up most of the time spent in code school. It is a relatively slow and sometimes tedious process. Several experimental studies were directed toward improving the training given during this period. These studies covered four general fields: an evaluation of a method comparable to the code-voice method described in the pre- ceding section but adapted to use at higher code speeds; studies of the most effective dis- tribution of practice; the development and ex- perimental test of methods of introducing variety into practice materials; and an analysis of the errors made by student operators. The Call-Back Method The success of the code-voice method in the early stage of training suggested that an adap- tation of it might be advantageous at more advanced stages. The call-back method was therefore developed for use with practice ma- terials which consisted of continuous messages or series of code groups. The call-back method systematized the fre- quent practice of reading to the students a series of characters that they had just copied and allowing them to correct their own copy. Daily records of the number of errors made in 100 consecutive characters allowed each man to plot his own learning curve. The method was evaluated experimentally by comparing the progress of two groups of sub- jects, one taught by the call-back method and the other by a method as similar as possible in all respects except for the omission of the call- back feature.12 The call-back procedure, at levels of 7 wpm and up, did not increase the student’s rate of progress. Neither did it enable him to achieve a higher level of final proficiency. On the other hand, it did not hamper achievement. The oc- casional use of the call-back method as a variety device to reduce the monotony of receiving prac- tice and as a means of satisfying those who want some knowledge of their progress was recommended. Distribution of Practice in Code Learning In an effort to force students to attain higher operating speeds, men in the code school at Camp Crowder were given 7 hours a day of code instruction. It seemed likely that this schedule was accomplishing no more than could be accomplished in fewer hours per day. In a variety of previous studies it had been shown that increasing the daily hours of drill or work beyond an optimum does not lead to increased learning or increased output. An experimental comparison of the effect of different numbers of hours per day for code practice was there- fore agreed upon,14 Several preliminary experiments were con- ducted with small groups. Five hours a day appeared to be at least as good as 7. Four hours per day was as good as 5. Three hours per day showed only a slight loss. Two hours per day could not be tried out because that schedule would have violated an order specifying the minimum total number of hours of code train- ing required of the men. On the basis of these preliminary experiments the project and the radio training section agreed to study a 4-hour-a-day schedule more inten- sively. The men were actually available for instruc- 110 TRAINING RADIO OPERATORS tion for 8 weeks. In the normal schedule they spent 7 hours a day, 5 days a week, and 4 hours a day on Saturday for 5 of these weeks in code school. The final 3 weeks were devoted to other topics. Code learning in a group of 355 men working on this schedule was compared with that of 165 men who had their training spread over the entire 8 weeks on a 4 hours per day, 6 days a week schedule. The same instructors were used with the two groups; the same cri- teria of mastery were employed at the different parison, the superiority of the shorter daily practice period is even more obvious. This com- parison is given in Table 2. Table 2. Code speeds attained hours of daily drill. with 4 and with 7 Highest speed passed Seven hours per day for 5 weeks No. of men Per cent Four hours per day for 8 weeks No. of men Per cent 7 wpm 10 wpm 12 wpm 15 wpm 18 wpm 20 wpm Total 15 4.23 58 16.34 151 42.53 112 31.55 16 4.51 3 0.84 355 100.00 3 1.82 22 13.33 71 43.03 36 21.82 33 20.00 165 100.00 X2 = 155.11 £ = .97 Median speed at end of training 13 wpm 15 — 16 wpm The conclusion from this experiment was un- questionable: 4 hours per day of drill was just as effective as 7 hours per day. The schedule was changed at Camp Crowder to put this find- ing into effect. The change had desirable by- products in addition to the principal advantage of saving time. Fewer men failed and morale immediately improved. A comparison was made of two code-teaching schedules as they affected the progress of 165 men in three successive classes throughout a 10-week period of observation.13 One group (74 men), the massed-practice group, was taught code during the first 4 hours of each 7-hour school day; another group (91 men), the spaced- practice group, was given instruction during the first, second, fourth, and seventh hours of the day. Progress was measured in terms of (1) hours required to pass successive code speeds, (2) number of men passing each speed, (3) performance on the Code Receiving Test (see Section 12.2.4) at the end of the training period, and (4) the highest speed at which students were able to receive their own sending (their operating speeds). Results of these comparisons indicated no significant difference in the progress of the two groups. In so far as the arrangement of Figure 1. Comparison of 4 and 7 hours per day of code practice in terms of mean receiving speed attained. A “day of practice” meant 4 hours of drill for one group and 7 hours of drill for the other. speed levels studied; and other training condi- tions were kept as alike as possible. The men of both groups were without previous code ex- perience. No significant difference existed be- tween the two groups in code aptitude or general intelligence as indicated by Radio Operator Aptitude and Army General Classification Test scores. The results are shown in Figure 1. This fig- ure shows clearly that 4 hours a day of practice produced just as rapid learning as did 7 hours a day. If the total training period (5 weeks for the 7-hour-a-day group and 8 weeks for the 4-hour-a-day group) is used as a basis of com- TRAINING MEN TO RECEIVE INTERNATIONAL MORSE CODE 111 4 code hours within a school day was concerned, one schedule was as effective as the other. Variety of Practice Materials in Code Drill A general complaint from students in radio code schools was that their many daily hours of copying code became very monotonous. This complaint was especially acute in the stretch of weeks after the code alphabet had been mas- tered and before the student was competent to handle real communications equipment and to work on simulated nets. The existence of mo- notony was recognized by officers and instruc- tors who persistently asked for ideas for re- ducing the deadly sameness of hour-by-hour copying of code. It was common in most code schools, during the middle weeks of the course, for the students to copy for several periods a day automatically transmitted code. The automatic transmissions were occasionally relieved by the instructor’s hand sending. In some schools the content of the drill was almost all of one kind; in others, the tapes were composed of messages more or less varied in nature. The pitch of the signals was usually the same, week in and week out. The instructors’ duties were chiefly those of room clerks, supervisors of discipline, and test ad- ministrators. Some schools had tried a few in- novations aimed to enliven the course. Some had short intervals of music or broadcasts of news and information in the middle of the periods. But no school, as far as the project could learn, had set up a comprehensive, sys- tematic program aimed to reduce monotony by employing a variety of activities which were designed to induce positive motivation to learn- ing. Therefore, the project staff decided to try out the effects of a program in which the activi- ties in each code period were continuously and systematically varied.8 It was agreed that in a 50-minute code period there should never be more than 15 minutes of continuous activity on any one particular kind of material and that usually there should be at least four separate activity-units in each period. Four principles governed the selection of ac- tivities to be included in the program. 1. The activities should pertain to code. Mo- notony was not to be relieved by stories, music, or other noncode activities. 2. The activities were to be adapted to large classes and mass education. 3. Each of the activities included was to be educationally useful in and of itself. 4. The activities should be made as interest- ing as possible. A variety of activities meeting these criteria can be found in reference 9. The value of this variety, or antimonotony, program was tested in two code schools. In each school the progress of students taught by the methods regularly employed was compared with the progress of students taught by the variety techniques. The experiment was a rigor- ous test of the variety technique, for the two schools chosen for the study were the two in which students normally made faster progress than in any other schools with which the project had had experience. Results differed in the two schools. In one, normally the best school known to the project, the classes taught by the regular procedures progressed just as rapidly as did those taught by the variety methods. The program of this school already included more variety than usual; the systematic addition of still more variety did not improve learning. In the other school, the second best normally, the variety classes learned more rapidly than the regular classes. From the eighth week on, the average man in the variety group could copy correctly from IV2 to 2 groups per minute [gpm] faster than the average man in the regular classes. The results from both schools and including only men in the two groups who were matched initially in terms of code aptitude were com- bined, The results obtained are shown in Table 3. The variety of activities led to better learning.8 In addition to improvement in code speed, several other positive values were reported from the variety experiment. Both students and instructors were better motivated. Both had to work harder, but the work was more interest- ing, The bored, lackadaisical attitude of the students and instructors gave way to one of alert interest. 112 TRAINING RADIO OPERATORS Use of Hand-Sent Clear Text as Practice Material Most practice material was in the form of five-letter code groups such as BDXMC, IATNB. Most of the practice material was machine sent. In a few schools the men were given occasional practice in copying hand-sent material. Oc- casionally part of the practice material was in plain English text, for example from news- papers, bulletins, or announcements. There was considerable disagreement over in short sentences. After each item was trans- mitted, the instructor read it aloud to his stu- dents in order that they might correct their copy. The amount of material transmitted in a practice hour varied with the speed level em- ployed ; at the 15 gpm level, about a page and a half of typewritten text was used. The data summarized in Table 4 justified sev- eral generalizations, 1. Under the schedule employed in this school, a daily hour of receiving hand-sent clear text, at any speed or speeds, did not retard progress in learning to copy cipher material. 2. The use of hand-sent clear text may actu- ally accelerate the progress of students in re- ceiving cipher. This was definitely true of those students who received clear text at both 12 and 15 gpm. 3. When clear text was used at one speed only, there was no significant difference be- tween the groups; when used at two speeds. Table 3. Comparison of code learning under normal and variety schedules. Regular classes, Variety classes, Test number gpm gpm 1 4.6 5.0 2 6.4 6.5 3 7.5 8.0 4 8.5 9.4 5 10.0 11.0 Table 4. Effect of one hour a day of practice with plain English on ability to receive cipher material. Receiving speeds Control group (cipher material only) Mean cr Hours of practice work Experimental group Differ- (1 hour per day of ences plain English) between means Mean