Chapter 14: Inspection and Statistical Quality Control
Inspection of Ordnance matériel in World War II appears at first glance to have been a fairly cut-and-dried affair. Weapons, rounds of ammunition, and vehicles produced by industry were accepted if they conformed to drawings and specifications, and rejected if they failed to conform. The casual observer assumed that the inspector had merely to examine each item to discover obvious surface defects, make specific measurements, and perhaps check on the weight or other physical characteristics of the item. But Ordnance inspection in World War II was far more complicated than this description suggests. Though some inspection was routine in nature, much of it was complex, difficult, and troublesome. Few Ordnance functions raised as many problems as did inspection; few were as important to the safety and welfare of troops in the field.
Ordnance inspection differed from standard commercial inspection chiefly because of the use to which Ordnance matériel was put. Drawings and specifications for guns and ammunition called for closer tolerances than most commercial work, for a weapon that exploded or failed to fire in an emergency might cause loss of life among American troops, result in a tactical setback, and have a bad effect on the morale of troops. Yet all inspection standards had to be geared to industry’s ability to manufacture matériel in quantity to close tolerances. Of necessity, they represented a compromise between the ideal and the practical.1
Each class of Ordnance matériel had its own inspection procedures and requirements. Rifles and machine guns were visually inspected, measured with a variety of instruments, and then given the test of actual firing at a small range near the plant. Spare parts for small arms were inspected too percent for conformance to specifications. Tanks, trucks, and artillery pieces could be visually inspected, measured, and put through their paces at a proving ground. Inspection of fire control instruments, particularly optical elements, called for special techniques because of the great precision required in their assembly. Ammunition, because of its explosive nature, was in a class by itself. There were many weighing and measuring tests for ammunition, but the only sure way to find out whether a round would function properly was to fire it and thus destroy it. Small arms ammunition required test firing of small samples from each lot, usually at a range near the plant; samples of artillery ammunition were test fired at one of the Ordnance proving grounds.
The broad heading of inspection embraced several different types of activity. “Surveillance” inspection was applied to matériel in storage, such as ammunition, that was subject to deterioration with the passage of time. “In-process” inspection was applied at various stages along the production line to check on processes; it was normally the function of the contractor, not of Ordnance. Another type, known as “screening” inspection, permitted acceptable items to go through and culled out the nonacceptable. Ordnance sometimes applied screening inspection to critical items but this type of inspection was normally the function of the contractor. Ordnance inspectors concerned themselves primarily with “acceptance” inspection, the final acceptance or rejection of matériel offered by the contractor in fulfillment of his contract. But acceptance inspection sometimes called for inspection of matériel during manufacture, before final assembly. Certain gears of a transmission, for example, had to be inspected before the transmission was assembled and placed in a vehicle. This type was really acceptance inspection but was occasionally referred to as “in-process” inspection.2
In theory, every piece of ordnance could easily be classed as acceptable or not acceptable by determining whether or not it conformed to drawings and specifications. But inspectors encountered all sorts of borderline cases that called for close study by engineers before final acceptance or rejection. Good judgment had to be mixed with engineering knowledge, familiarity with production processes, and an understanding of the functioning of the end item. No production line could turn out quantities of absolutely identical and acceptable items; there was always some variation, though it might amount to only one thousandth of an inch. Permissible variations shown on Ordnance drawings and specifications in the form of “tolerances” were usually on the conservative side. From experience and training, Ordnance inspectors knew that matériel that deviated from the tolerances set forth in the drawings and specifications might function perfectly—or might fail at a critical moment. Ordnance did not wish to reject serviceable matériel on the narrowly legalistic ground that it did not conform to the letter of the requirements; nor did it wish to take too liberal a view and run the risk of accepting matériel that might prove unserviceable, perhaps even dangerous, when issued to troops. Writing about this problem as it concerned artillery ammunition, one inspection specialist summed the matter up as follows:
Conformance to a design implies that there exist arbitrary limits to variations in dimensions, in finishes, in materials, and so on. This is true, in a legal sense, since drawings and specifications prescribe such limits. From an engineering point of view, however, there generally do not exist sharp boundaries between good and bad. For most dimensions, and for most other prescribed properties, an increase in variation means either a decrease in effectiveness of the ammunition or an increase in probability of obtaining a malfunction. In most cases, the effect on functioning is very gradual, so that a very considerable variation may exist before the results become apparent in the small sample subjected to proving ground test.3
Once matériel passed inspection it was marked with the Ordnance escutcheon, commonly referred to as “the crossed cannon and bomb in circle,” using a
stencil, stamp, seal, or tag. For some classes of items, identifying lot numbers or serial numbers served as evidence of inspection. When an inspector gave matériel only provisional acceptance he marked it with the Ordnance insigne, the flaming bomb. Artillery weapons that underwent proof testing were marked with the letters “P.A.” (for small arms, “P” or a prick punch mark) followed by the initials of the proving ground. Matériel that failed to pass inspection was marked with a stamp, die, red rejection seal, or tag consisting of a large “X” in a circle with the words “Ordnance Rejected.” All such matériel was carefully segregated to prevent its entering the production line until reworked to meet Ordnance standards, or until the Office Chief of Ordnance granted a waiver for it or decided to scrap it.
Ordnance did not inspect everything that it procured from industry. In some instances it accepted products on the strength of a contractor’s certificate that they met the specifications. The contracting officer might accept such certificates in lieu of inspection when, for example, the product was a standard commercial item and past performance of the contractor had been particularly good. Certification not only helped to conserve inspection manpower but also promoted mutual respect and understanding between industry and the Ordnance Department.
Inspection Manuals
During the 1920s and early 1930s the Ordnance Department procured so little matériel that inspection was not a major problem. Practically every item that came off the production line was painstakingly examined, measured, and weighed. Each arsenal provided inspection service on matériel that fell within its domain; each had on its staff inspectors with long experience in their work. As early as 1922 the Chief of Ordnance, Maj. Gen. Clarence C. Williams, had recommended that the arsenals keep alive the art of inspection and be prepared to train inspectors in a future emergency when the districts would take over responsibility for inspection as part of their procurement function. As one of its preparedness activities, Ordnance in 1935 drafted a General Inspection Manual and circulated it to all the arsenals and district offices, followed three years later by a revised edition and by manuals on specific classes of matériel.
The 1938 manual continued to serve as the basic general guide for Ordnance inspection until 1945. One of the most important principles it set forth was that in-process inspection was the contractor’s responsibility and that Ordnance inspectors should, wherever possible, be limited to inspection of end items. The manual set high standards of conduct for Ordnance inspectors because they were the personal representatives of the Ordnance Department in dealing with industry. In the eyes of workers in the plants, inspectors were “the government.” “It is the desire of the Ordnance Department,” the manual stated, “to have its inspectors ... respected for their integrity, ability, impartiality, tact, thoroughness, and prompt and business-like methods of conducting inspections.” Because they were usually the only government representatives in plants, inspectors were often called upon to perform many services not related to inspection, such as looking after government equipment in the plant or helping to solve production problems.
A few inspection manuals for specific items such as artillery shells and cartridge
cases were in use. They described the items, told how they were manufactured, and listed possible defects that the inspector was to look for. Defects were sometimes classed as critical, major, or minor, depending upon their importance to the proper functioning of the item. For some critical characteristics the manuals required too percent inspection; for others they stated that inspection of a certain percentage of items would be sufficient. As a rule, the manuals provided that the percentage of items inspected could be reduced if quality remained consistently high. Thus, during this early period, Ordnance was applying some of the basic principles of statistical quality control by using sampling techniques, classifying defects according to their importance, and gearing inspection to known quality level.
Recruiting and Training Inspectors
The Educational Orders Act of 1938 marked the revival of inspection activities in the district offices. Though the Chief of Ordnance retained in his hands close control over all educational orders, he delegated to the districts responsibility for inspecting the final product. The Boston district hired its first inspector in June 1938 and assigned him the task of bringing up-to-date the district’s file of specifications and drawings. He also inspected whatever matériel was procured and, as time permitted, made plant surveys.4 Other districts followed suit and most were able to recruit competent men to handle inspection of the small quantities of matériel procured under educational orders. Most districts, in fact, hired more inspectors than they needed at the start. The work called for a good deal of traveling, for inspectors usually went to contractors’ plants to inspect matériel before it was packed for shipment. Stretching meager district funds to pay for travel of inspectors caused many headaches for district executive officers during the 1938-40 period. Later, when contractors achieved steady output on regular production orders, the districts assigned one or more inspectors to each plant on a full-time basis.
As the inspection work load mounted during 1938 and 1939 the districts appealed to the arsenals for help in supplying qualified inspectors. Nearly every district obtained one or two arsenal inspectors, but the arsenal commanders, faced with mounting work loads of their own, were reluctant to release more. The other source of supply, recruitment through Civil Service, gradually dried up as industry absorbed more and more workers at rates of pay higher than those offered inspectors under Civil Service.5 At times the districts experienced exasperating delays, some stemming from pure red tape, in obtaining approval by the Civil Service Commission and the Chief of Ordnance of employees hired as inspectors.6 To help meet such problems, the qualifications for inspectors were lowered, civilian schools were encouraged to offer courses in inspection techniques, and in June 1940 the Chief of Ordnance directed each district to send several of its most promising inspectors to one or another of the arsenals
for a 6-weeks’ training course.7 Each arsenal instructed the trainees assigned to it on the items it normally produced, and repeated the courses as new groups of trainees arrived. Meanwhile each district assigned a few of its reserve officers to inspection work. Some took inactive duty training in inspection methods; others spent their terms of active duty in manufacturing plants observing the work of resident inspectors.
The districts were not altogether satisfied with the training their inspectors received at the arsenals. Some of the arsenals gave excellent instruction; others merely turned the students loose in the shops to learn what they could.8 The main complaint was that the arsenals encouraged inspectors to use their own judgment in dealing with borderline cases. The districts felt that uniformity of inspection could never be achieved if inspectors were permitted to use their own judgment in accepting matériel that did not comply with specifications and drawings. District officials wished inspection standards to be as uniform as possible so that no contractor could complain that his products were rejected while similar products submitted by a competitor were accepted. The districts were keenly aware of the fact that most of their inspectors had too little experience to be counted on for exercise of good judgment on engineering problems.
During 1941 many of the Ordnance districts arranged with local schools or colleges to offer prospective inspectors training in elementary mathematics, physics, blueprint reading, mechanical drawing, machine shop practice, and the use of measuring instruments. During their time of study in these courses the trainees held the rank of Under Inspector, CAF-2, received a salary of $1440 per year, and upon graduation became Junior Inspectors, CAF-3, earning $1620 per year. Trainees received pay during their schooling, because experience had shown that most of those who attended courses without pay took jobs in industry instead of working for Ordnance.
In spite of low salaries and other problems, the districts managed to recruit inspectors rapidly during the defense period. In New York, for example, the number of civilian inspectors jumped from 5 at the end of 1939 to 492 in December 1941; in the latter month the district had 54 reserve officers on active duty and had under inspection roughly 800 prime contracts and 1,000 subcontracts.9 At the end of 1941, inspectors accounted for about three-fourths of all civilians employed in each district. Most were in the lower Civil Service brackets; only a few were as high as CAF-9. Though they were technicians and should have been considered as Wage Board or subprofessional personnel, nearly all had CAF (clerical, administrative, and fiscal) ratings. As selective service took more men, women were hired to replace them. The turnover among inspectors was appalling. “If it were possible to secure higher grade inspectors,” the Philadelphia Ordnance District reported, “it would be possible to handle the work with less personnel. This would lead to higher average standards and fewer
rejections ...”10 In most districts a commissioned officer directed the inspection staff and served as adviser on inspection to the district chief.11
General Somers’ Role
Because of the growing importance of inspection, General Harris in July 1941 added Brig. Gen. Richard H. Somers to his staff in the Industrial Service as assistant chief for inspection. He assigned to General Somers responsibility for coordinating the inspection activities of the commodity branches, supervising acceptance testing at the proving grounds, and advising on inspection policies.12 But responsibility for production, both quantity and quality, continued to rest with the commodity branches and the district offices. General Somers was to be consulted on proposed changes in specifications that had a bearing on inspection, but he had no overriding authority to enforce inspection standards. This basic arrangement continued throughout the war, though General Somers retired in 1942 and his duties were assigned to the Production Service Branch of the Industrial Division.
In spite of some brave talk about making the inspection staff the independent guardian of quality, it never achieved true independence either in the Office Chief of Ordnance or in the districts. General Campbell in the summer of 1942 delegated inspection responsibility to the matériel operating divisions on the theory that the same officials should he held accountable for both quantity and quality. Inspectors in the districts felt that, because commodity branch chiefs gained recognition by meeting production schedules, pressure to boost production sometimes contributed to deterioration in quality. Branch chiefs re plied that quality was not absolute but relative. They contended that Ordnance had to deal with hard practical realities and that its objective was to procure the best matériel possible in the quantities required by the Army in the time available.13
There was, further, a diversity of practice among the district offices. Though all the districts procuring a given item used the same drawings and specifications there was no enforced uniformity among them on inspection procedures or organization. Early in the defense period the districts complained that some of the drawings and specifications issued by the arsenals were not up-to-date and did not show the latest changes in design. But, as time wore on, these discrepancies were corrected. When the ASF survey team headed by Dr. Luther Gulick visited Cincinnati in April 1942 it found that contractors were generally satisfied that government inspection was both fair and necessary.14
Inspection Gages
As mentioned in an earlier chapter, during the 1930s nine of the Ordnance districts set up gage laboratories, usually at universities, where acceptance gages could be checked for accuracy. In 1940 and 1941 the remaining districts and all the arsenals established similar laboratories, raising the total to nineteen.15 As the Ordnance inspection force grew during the defense period, and as more and more items of equipment went into production, the need for gages and gage laboratories steadily rose. A single example illustrates the scope of the problem: one type of fuse for the 75-mm. shell consisted of some thirty-five metal parts and required over two hundred different gages for its inspection.
Ordnance met the early demands by using the half million old gages in storage. In 1938 and 1939 the Ordnance Gage Section vigorously pushed efforts to design gages for all items that were reasonably sure of going into production. In July 1940 approximately $2.5 million was made available to start production of new gages well ahead of the signing of major procurement contracts. In October 1940 President Roosevelt approved a $4 million program to enable gage manufacturers who had already expanded their plants with their own funds to increase their production capacity still further.16 As the Ordnance Department was the agency with the greatest need for gages it was placed in charge of the expansion of gage capacity for all elements of the Army and Navy. In 1941 Ordnance gave another boost to gage production by purchasing machine tools and leasing them to gage manufacturers.17
These timely steps averted a critical shortage of gages that might otherwise have developed. But Ordnance occasionally had to resort to temporary expedients. By the end of the defense period, when production volume was rising fast, Ordnance inspectors sometimes had to borrow gages from contractors whose products they were inspecting. Whenever this was done the borrowed gages were first sent to a district gage laboratory to be checked for accuracy. Later, as Ordnance gage procurement caught up with demand, inspectors were required to submit their gages to a laboratory at intervals for checking. As an added safeguard, roving teams of gage checkers visited plants, examined Ordnance gages, and checked on their use.18
Proving Grounds
From its establishment in World War I, Aberdeen Proving Ground in Maryland had been the principal Ordnance center for proof firing of weapons and ammunition, but it was not capable of handling the heavy, diversified work load anticipated in 1940-41. Ordnance had established a second test center during World War I, Erie Proving Ground adjacent to
Camp Perry, Ohio, but had converted it in 1920 from a proving ground to a Field Service storage depot. Late in 1940 Ordnance decided to re-establish Erie as a proving ground for acceptance testing of guns, carriages, mounts, recoil mechanisms, and armor plate. At about the same time Ordnance selected a 50,000-acre site for a new proving ground near Madison, Indiana, in the heart of the ammunition-producing area. Named Jefferson, it was to proof fire all types of ammunition, ranging from small 20-MM. rounds up to heavy 240-mm. shells, from hand grenades to giant bombs. The first shot was fired at Jefferson on to May 1941, just a short time before construction began on another 50,000-acre proving ground near Hope, Arkansas. Named Southwestern, it had the mission of proof testing primers, fuses, boosters, cartridge cases, propellants, bombs, pyrotechnics, and, late in the war, rockets. It fired its first shot on New Year’s Day 1942. As these three new proving grounds came into service Aberdeen did less acceptance testing and devoted more time to research and development tests.19
Statistical Quality Control
One of the most significant developments in Ordnance inspection during World War II was the use of statistical sampling techniques under the name of “quality control.” The origins of this practice in industry are usually traced back to the year 1924 when Dr. Walter A. Shewhart and his associates in the Bell Telephone System began to apply statistical analysis to the inspection of large numbers of production items. Seven years later Dr. Shewhart set forth the principles of what some writers called “a new science” in his book, The Economic Control of Quality of Manufactured Product (New York: D. Van Nostrand Company, 1931). But progress in the industrial application of the principles was slow.20
Simon’s Pioneering Work
The first Ordnance Department experience with ‘the subject came in 1934 when 1st Lt. Leslie E. Simon at Picatinny Arsenal began to study and apply Dr. Shewhart’s work.. Picatinny was a logical choice because it endeavored to make large numbers of rounds of ammunition as nearly identical as possible. As Dr. Shewhart’s home was conveniently located in nearby Mountain Lakes, N.J., Lieutenant Simon soon became acquainted with the “father” of quality control and interested him in becoming a consultant to the arsenal. Under Simon’s direction the arsenal drew up and published a practical pamphlet for shop use called Instructions for Control of Quality Thru Percentage Inspection.21 In spite of the gains registered in the early tests, Picatinny’s interest in the subject waned after Simon’s transfer in 1937, and was not aroused again until the danger of war brought the need for mass production of munitions. In 1941, Simon, now a major and assistant director of the Ballistic Research Laboratory at Aberdeen Proving
Laboratory foreman loading 37-mm. antitank gun for test shot at Erie Proving Ground, July 1941
Ground, published some of his thoughts on the matter in book form under the title, An Engineer’s Manual of Statistical Methods, which soon found its way into college classrooms as well as into industrial plants.
Simon illustrated the nature of the problem faced by inspectors dealing with mass production items by pointing out that small samples taken at random from a moderately defective lot of items would not accurately reflect the quality of the lot; they would consistently show a lower proportion of defectives than the average for the entire lot. If, for example, Ordnance set up a requirement that each lot of one hundred items would be accepted when a sample of ten items selected at random from the lot contained not more than one defective item, it would in fact be accepting lots that averaged a good deal more than 10 percent defective. A commonly applied remedy for this condition was to require that the inspection sample contain no defectives at all. But Simon found this remedy far from satisfactory. “It is true,” he wrote, “that it somewhat reduces the chances of acceptance of poor quality; but its penalties fall both on the just and the unjust, and it results in rather high rejections of relatively good quality.”22
Simon’s answer to the problem was not to reject the sampling technique but to buttress it with other evidence by keeping records that would tell the history of a given production run, would indicate its expected quality level, would sound the alarm when any variation in quality occurred, and would tell the producer when
and where to look for trouble. The practical man, Simon noted, does not confine himself to the evidence of the sample, but instinctively considers everything he knows about the production process. This was essentially the line of thought earlier developed by Dr. Shewhart; it became the basis for the application of statistical sampling or “quality control” in Ordnance World War II inspection.23
The Edwards Survey
Soon after the Japanese attack on Pearl Harbor, General Somers decided to explore the possibility of making further use of statistical sampling techniques in Ordnance acceptance inspection. Their use was growing in industry and they were being applied for special purposes in the Ammunition Division, at the arsenals, and at Aberdeen Proving Ground. According to their proponents, these techniques yielded better control of quality, required fewer inspectors, and reduced the amount of matériel destroyed in testing. The importance of this latter item is suggested by the fact that in 1942, Ordnance was shooting up enough armor plate in ballistic tests to make approximately thirty medium tanks a month.24 One of the first steps taken by General Somers was to enlist the services of George D. Edwards, director of quality assurance of the Bell Telephone Laboratories, as consultant to the Ordnance Department. Edwards was asked to survey existing Ordnance inspection methods and recommend ways of increasing their effectiveness by using the newest techniques of statistical quality control.25
The Edwards survey revealed that there was much room for the use of quality control methods in the acceptance inspection of matériel produced for Ordnance by industry. Lieutenant Simon’s work at Picatinny had dealt mainly with in-process inspection of matériel being manufactured at the arsenal, and the bulk of all Ordnance inspection during the interwar years had been of this type. But by 1942 the situation had changed. The arsenals were producing only a small percentage of the Army’s munitions. Industry was the chief source of new matériel, and the Ordnance task was that of acceptance inspection of industry’s products. The Edwards survey also revealed that few persons in Ordnance, whether military or civilian, knew anything about quality control, and many were opposed to its adoption in their fields of inspection.26
Edwards took an eminently practical approach to the inspection problem. He recognized that no feasible plan of inspection would guarantee rejection of every defective item. The best that any inspection system could hope for was reduction to a minimum of the risk of accepting defective items without unduly holding up production. “The hard facts are,” he wrote, “that we must have ordnance and we must accept and get along with
ordnance of the highest quality which can be produced in the quantities which we must have under present conditions.”27 The problem was to determine a level of acceptable quality and then draw up a statistical sampling plan that would pass matériel of the desired quality and would immediately sound an alarm when quality fell below that level.
The Quality Control Campaign, 1942–43
The plan of campaign that resulted from the Edwards report called for two distinct attacks on the problem. Ordnance was to select one or more suitable types of matériel and apply quality control principles to their inspection. This would provide working examples of the system. The other attack was to develop a training program within Ordnance to teach the basic principles of quality control.
Ballistic testing of armor plate was selected as one area for the application of quality control principles, beginning in July 1942. The method involved plotting on a control chart the results of tests on armor submitted by each manufacturer. If the chart for a given manufacturer indicated that he was consistently producing acceptable armor Ordnance reduced the number of tests made on his product by two-thirds. Testing continued at this reduced rate as long as the chart indicated that the manufacturer was properly controlling the quality of his production. But if the number of rejections exceeded a certain level the reduced inspection rate was discontinued, inspection went back to normal, and a search was made for the factor that had caused quality to decline. Ballistic testing was, of course, by no means the only way of assuring quality of armor. Laboratories at the contractors’ plants performed metallurgical tests; the manufacturers’ knowledge and past experience in steelmaking gave added assurance that their production methods were sound.
Beginning in September 1942 the Inspection Branch, under Edwards’ direction, conducted a series of 3-day conferences on statistical quality control. Some 220 officers and civilians from the district offices, arsenals, plants, works, and proving grounds, attended and gained at least a rudimentary knowledge of the basic principles.28 One of the main themes stressed at the training conferences was that in-process inspection was the contractor’s responsibility. The conference leaders demonstrated that sampling inspection, by accepting or rejecting large lots on the basis of small samples, would force contractors to screen out defective items before presenting a lot to Ordnance for acceptance. If contractors did not do so they would run the risk of having large lots rejected and sent back for screening.29 While the training conferences were being held the Industrial Division issued a directive to the effect that Ordnance inspectors should perform only acceptance inspection. It called for the elimination of all in-process inspection that had been provided to aid contractors in getting production
under way.30 By February 1943 the Industrial Division was able to report that over 400 inspection plans embodying quality control principles in the inspection of some 150 types of matériel were in effect. To push the program further, General Hayes in that same month directed the branches of the Industrial Division, the Tank-Automotive Center, and all the district offices to appoint an officer or civilian to be responsible for quality control.31
As a result of these and related efforts, the number of Ordnance inspectors per million dollars of accepted matériel dropped from forty in September 1942 to about nine in April 1943. While the value of accepted matériel rose from $500 million in September 1942 to $1400 million in April 1943, the total number of Ordnance inspectors dropped from twenty thousand to thirteen thousand. “Such a record,” wrote one inspection specialist, “can be attributed entirely to Ordnance contractors accepting the responsibility of producing satisfactory material before presenting it to Ordnance for acceptance.32 But there were other elements in the picture, too. The Army-wide drive to conserve manpower was in full swing during the winter of 1942-43; in some instances Ordnance reduced its inspection force at the cost of lowered quality. Some economies naturally resulted from volume production as well as from increased use of sampling methods. As manufacturers got into production and gained experience with Ordnance requirements the need for meticulous inspection of every item by Ordnance inspectors declined. The Boston district conserved inspection manpower by forming small teams of traveling inspectors to handle the work at plants that did not produce enough matériel to justify full-time resident inspectors.33
At no time did Ordnance tell its contractors exactly what inspection to perform. It gave advice when requested and made a practice of conferring with each, contractor on inspection matters as soon as his contract was signed. But it did not give its contractors detailed instructions on how to inspect. Had it done so Ordnance would have been morally bound to accept what the contractors turned out in accord with those instructions. Instead, Ordnance kept its hands free to accept or reject the finished items presented to it.34
The Trundle Report
While Ordnance was thus feeling its way in the quality control area, Army Service Forces decided to prepare an Inspection Manual as a means of simplifying and coordinating the inspection procedures of all the technical services. To this end it enlisted the services of the Trundle Engineering Company of Cleveland, Ohio, to survey existing practices and later prepare the manual.35 When they made their
report in the summer of 1943 the Trundle analysts observed that the inspection mission of the technical services had become “a stupendous and complex task.”36 This task had grown with bewildering speed from the peacetime year 1939, when the U.S. Army numbered about 188,000 men, to the second year of war when the strength of the Army exceeded 7,000,000 men and annual expenditures for munitions ran into the billions of dollars. “Haste inevitably contributes to confusion, friction and ineffectiveness,” the report continued. As a result it found that there was urgent need for improving, simplifying, and standardizing the inspection work of all the technical services.
The section of the report dealing with Ordnance criticized the districts for placing incompetent persons in responsible positions, for maintaining only loose control over inspectors, and for tolerating—or being ignorant of—”inefficient handling and duplication of forms in the field offices of the resident inspectors.” The report scored the lack of uniformity among the inspection practices of the Ordnance districts, pointing out that one district would accept such raw materials as steel, paint, grease, and oils without inspection or test while another district would go to great lengths to inspect and test the same material.
Considering all the technical services, and without special reference to Ordnance, the report concluded that government inspectors were doing too much in-process inspection. It did not cite examples of undesirable Ordnance in-process inspection, nor did it describe the history behind the situation that it criticized. As noted above, the demand for production had been so great during 1941-42 that Ordnance had tried to help speed output by placing inspectors in plants to perform inspection that the contractors would normally have performed themselves. But after October and November 1942, when the districts were ordered to stop this practice, the only type of so-called in-process inspection officially approved was the inspection of certain parts, such as gears in a crankcase, before they were assembled and became inaccessible.37
The report also concluded that none of the technical services was making enough use of statistical quality control, though the report did not spell out in detail how or where Ordnance was deficient. In view of the pioneering work the Ordnance Department had done in this area the Ordnance inspection staff felt that this criticism was not fully justified. Ordnance had pushed forward with the adoption of statistical quality control techniques during the preceding year at what was considered to be prudent speed. It had gone farther and faster than any of the other technical services and was steadily advancing at the time the Trundle survey was made. The Ordnance inspection staff felt that it deserved commendation for its achievements rather than censure for its shortcomings.
Commodity Groups
The real story of Ordnance inspection can be told only by dealing with individual groups of commodities handled by the matériel operating divisions of the
Industrial Service. Each group was different. Each had its own peculiar problems and each found its own solutions. The following accounts put the spotlight briefly on rifles, tank-automotive matériel, and fire control instruments as three fairly representative types.38
Rifles at Springfield Armory
The history of the inspection of rifles at Springfield Armory sheds a good deal of light on the difficulties Ordnance inspectors encountered during the war and the progress they made in improving quality. In peacetime, when Springfield turned out only small quantities of rifles and rifle parts, inspection was a slow and painstaking business. High quality craftsmanship was the order of the day. But with mounting requirements for M1 rifles in 1940 and 1941 the armory was called upon to expand its shops and turn out rifles by the million. Under these circumstances inspection had to take a back seat. Minor defects in parts were ignored if the rifle fired satisfactorily when tested. All rifles were test fired with one high-pressure proof round and twenty or more normal rounds. A small percentage from each lot underwent a 6,000-round endurance firing; and a few were disassembled and checked for interchangeability. But there was no insistence on rigid adherence of all parts to drawings. In 1944 Dr. Constance McLaughlin Green, armory historian, wrote as follows:
The drive to meet schedules had increased month by month in intensity to a point where standards of work had been somewhat undermined, from machine operator up through top inspection ranks. Rifles shipped out had, to be sure, always met function tests. ... But this was true in spite of the fact that separate components often failed to meet the gage requirements. Presumably the tolerances entered on the drawings were closer than functioning of the assembled parts demanded. Still it was far from an ideal situation for the plant that was to serve as the model for all small arms manufacturers.39
In January 1944, with pressure for production eased, Lt. Col. William Gallagher took charge of the Inspection Department and mapped a vigorous campaign to improve quality. Though no complaints of defective rifles had come in from the field, the armory determined to improve its product in every way possible. After thorough study of the problems involved, Colonel Gallagher outlined four major steps:
(1) Housecleaning in the manufacturing department. Tools, fixtures, and machines that had been continually in use during the months of heavy production were to be overhauled and put into the most perfect adjustment attainable.
(2) Floor inspection at every machine. Instead of inspection at the end of three or four operations, a system of floor inspection at every machine was to be inaugurated so that machines in need of resetting or new cutters would be promptly detected and serviced.
(3) Education. The need for raising quality and holding every part within prescribed tolerances was to be sold to everyone in the armory, from machine operators up to production engineers and chief
inspectors. All employees were to be taught that quality was just as important as quantity.
(4) Periodic checking of manufacturing gages. The general practice earlier had been to send gages back for checking only when a foreman or an inspector had reason to believe that they were worn. By setting up a schedule for checking each gage periodically it was hoped that faulty gages would be eliminated as causes of manufacturing inaccuracies.
The arsenal commander promptly approved Colonel Gallagher’s proposals. They went into effect during the first six months of 1944.40
At the same time the armory’s inspectors were divided into two groups: manufacturing inspectors and acceptance inspectors. Complying with the ASF Inspection Manual issued in March 1944, the armory put its production division in the same position as an Ordnance contractor by divorcing in-process inspection from acceptance inspection. All manufacturing operations and in-process inspection were to be performed under authority of the Manufacturing Department; the finished products were then to be turned over to the final inspection staff for acceptance or rejection. At the same time the final inspection staff began placing all components of the M1 rifle on a statistical sampling basis.
The over-all results of these two steps were found to be good, but quality still did not rise to the level desired by the Chief of Ordnance. During the early months of 1945 the armory inspection staff held weekly meetings to get to the bottom of the problem. The experts went over every component, studied its methods of manufacture, and examined its gages. But progress was slow and piecemeal with no major improvement before the end of the war.
Tank-Automotive Matériel
Ordnance inspection officials realized early in World War II that they could not apply traditional inspection procedures to products of the automotive industry. The very magnitude of the task was appalling even when Ordnance had only combat vehicles to consider, for each tank and gun motor carriage consisted of thousands of individual parts. After the transfer to Ordnance of transport vehicles in September 1942 the inspection job reached staggering proportions. “The complexity of parts as well as automotive sub-assemblies,” wrote one observer, “caused inspectors to throw up their hands at the practicability of any statistically and technically logical approach ....”41 The Ordnance inspection staff recognized that the procedures developed over the years for inspection of “shooting ordnance” would, if applied to automotive products, result in too much inspection and the wrong kind of inspection. There was, for example, no danger of explosion in the normal operation of trucks or tanks such as there was with weapons and ammunition. Nor was the process of manufacture unfamiliar to industry. The manufacture of military trucks was to a large extent the same as manufacture of civilian trucks; even tanks were made up in part of components similar to standard commercial items. “We are dealing with the largest and most responsible industrial
units,” wrote the chief of the Inspection Branch on OCO Detroit, “where no one in his right mind would endeavor to furnish substandard matériel for any reason whatsoever.”42 Though Ordnance felt that the manufacturers of tank-automotive matériel could be depended upon to produce acceptable matériel, it also realized that occasional lapses were inevitable and that a certain degree of inspection was necessary to protect the government’s interests. Inspection policies were kept broad and flexible to permit their application to a wide variety of manufacturers.
Standard Ordnance procedure called for a functional test of every complete vehicle before it was accepted. This test included operating the vehicle on the road and trying out major units such as winches, lift devices, and turret traversing mechanisms. The contractors conducted these tests, under the eyes of Ordnance inspectors, at small test areas adjacent to the plants. To supplement such tests the Ordnance inspection force selected a few vehicles from each assembly line for special testing at one of the Ordnance proving grounds. These “inspection control tests,” as they were called, proved to be valuable as checks on the manufacturers, as a means of revealing weaknesses in design, and as a test of packing procedures.
Subcontractors who manufactured engines, transmissions, axles, and other major components ran each of these items on a test stand before delivering it to the assembly plant. As a rule, this functional testing was done on a tzoo percent basis. Little use was made of statistical quality control procedures, except in ballistic testing of armor, inspection of tank track components, and acceptance of pneumatic tires and tubes. With combat vehicles the proof firing of gun mounts was an added means of assuring that vehicles would perform properly. Ordnance also conducted interchangeability tests at the manufacturers’ plants or at the proving grounds on components selected at random from process lines. These tests were spot checks only, for Ordnance did not insist upon complete interchangeability of automotive equipment. In peacetime the automotive industry had never achieved izoo percent interchangeability; it was generally understood that in using spare parts some slight fitting was necessary. Ordnance realized that it was not feasible, in time of manpower shortages and high production goals, to insist upon a degree of interchangeability never attained by industry.
Ordnance inspection of tank-automotive matériel was marked by great diversity. The basic policy provided that Ordnance would conduct inspection upon end items and as far back in the production chain as necessary to assure quality products. Tank-automotive production was characterized by assembly in the prime contractor’s plant of many complicated subassemblies such as engines, transmissions, and axles. No single type of inspection would fit all these components. With some, materials and heat-treating controls were the essence of quality. With others, dimensional characteristics were the keys to proper performance. With still others there were simple operating tests that gave adequate assurance of quality. Most of the Ordnance tank - automotive contractors had enviable records in industry for quality production, but some were newcomers to the business who had little previous experience in making the parts needed by
Ordnance. Inspection requirements had to be tailored to fit the needs of each case.
Fire Control Instruments
Binoculars, telescopes, directors, and other fire control matériel ranked as perhaps the most difficult class of items for Ordnance inspection. They presented a wide variety of problems, for they included mechanical, electrical, optical, hydraulic, and electronic instruments. Production of good optical glass required careful control; machined parts of fire control instruments had to be held to close tolerances; and assembly of optical elements called for meticulous accuracy. Evaluation of the importance of scratches, pits, and stain on surfaces of optical elements was largely a matter of experienced judgment, as was the determination of allowable distortion in the glass. Dependence on human judgment led to innumerable difficulties. Added to these problems was the need to use inexperienced inspectors and to meet constant demands for speedy production.
“I think no one would willingly sacrifice or adulterate the quality of Ordnance supplies going to the fighting forces,” wrote the chief of the Fire Control Sub-Office, Col. Gordon B. Welch, in March 1945. “However, that has to be tempered with judgment when the attainment of high quality prevents the fighting forces from having anything at all. I ... have never hesitated to lower the quality in particular cases when it was necessary to meet our objectives.”43 As a result of all these circumstances, Ordnance accepted far too much unsatisfactory matériel in the early war period. American and British troops in North Africa in 1942-43 sent back a stream of complaints about defective fire control instruments.44
Reports that defective matériel had been issued to troops shocked the Chief of Ordnance and his staff. To correct the situation the Artillery Division strengthened the Inspection Section of the Fire Control Sub-Office at Frankford and set up new procedures calling for prompt action on reports of defective matériel. The chief of the sub-office wrote official letters to all the district offices and followed them up with a personal appeal to each district chief to stop the acceptance of substandard matériel. The Inspection Section pursued a vigorous program of interchangeability tests coupled with investigation of all deficiencies. As measured by these tests and by the number of defective matériel reports that came in, the quality of fire control instruments rose steadily from August 1943 to the end of the war.45
Some reports of defective matériel, thought to be caused solely by inadequate inspection, proved to be due wholly or in part to engineering design. With binoculars, for example, Ordnance received many reports that moisture and dirt had got into the instruments. As inspection standards for binoculars were strict, the reports were puzzling. Even after Ordnance began packing the instruments in vaporproof barriers with silica gel the reports persisted. Finally, a study of a large number of defective binoculars at Augusta Arsenal revealed that the so-called dirt within the binoculars was the result of a chemical
action on the reticle cell, which was made of secondary aluminum. Investigation also proved that after the binoculars underwent the rain test at the plant they contained minute quantities of water that escaped detection by the inspection measures then in use. Once these facts were brought to light, corrective measures were taken, and reports of defective binoculars dropped almost to zero.46
Most fire control instruments did not lend themselves to inspection by statistical sampling, chiefly because they were produced in small numbers and did not involve a high volume of repetitive operations. As a matter of policy the Fire Control Sub-Office made little use of quality control. Along with artillery weapons themselves, fire control instruments ranked lowest in the Ordnance list of items inspected by statistical quality control methods.47
All things considered, the Ordnance record on inspection was checkered. With some items there was always a gap between the quality prescribed by the drawings and specifications and the quality of the matériel actually accepted. Ammunition, for example, was produced in huge quantities and won an enviable reputation for quality and reliability. But it was not perfect. Even late in the war, after more than two years of steady production, one investigator who checked on the manufacture of metal parts for fuses reported that, “the quality of matériel being accepted by Ordnance inspectors is nowhere near that which has been established as acceptable.”48 The same was true in greater or less degree of weapons, both small arms and artillery, and of trucks and combat vehicles. They functioned well, as a rule, but they did not always comply exactly with specifications and drawings. In many instances there was reason for excusing noncompliance on the ground that the Ordnance tolerances were too strict. But in other cases it may safely be assumed that, had the manufacturers (including the arsenals) kept within the limits set by Ordnance, the end product would have functioned better, lasted longer, or been more reliable.
The chief reason for failure to maintain the highest quality was pressure to get out production. Ordnance realized that in time of war the overriding requirement was for good munitions in huge quantities, not perfect munitions in small quantities. The highest standards of precision manufacture were impossible of attainment in a war economy where skilled workers were scarce, the demand for speedy production was intense, and machine tools ran every day with very little time out for maintenance. Production managers looked upon the rejection of material by inspectors as something on a par with throwing a monkey wrench into the machinery. It caused loss of time, labor, and materials, and it played havoc with scheduling. Coupled with this was the fact that contractors could sometimes prove that Ordnance tolerances were unnecessarily close, or that inspectors were rejecting matériel for trivial defects. All these elements conspired to make the maintenance of high standards of quality a very difficult task. The result was a compromise between the ideal and the practical.
The most notable new development in Ordnance inspection was the introduction of statistical quality control. Ordnance took pride in being a pioneer in the use of quality control techniques. How successful its efforts Were is hard to measure, for the techniques were not applied equally to all types of matériel. But the following summary statement drafted early in 1943 by the wartime chief of the Quality Control Unit in the Ammunition Branch appears to be close to the truth:
Ordnance inspection is becoming more efficient, Government inspectors are beginning to accept and reject on a more rational and standardized basis, rule of thumb is being eliminated, quality of accepted material is improving, and Ordnance inspection personnel are, in general, being reduced. At the same time, the responsibility of manufacturers to submit only material of satisfactory quality for acceptance is being more definitely crystallized, and greater cooperation is being obtained.49