Some Observations On The Failure
Of U.S. Model 1903 Rifle Receivers
Joseph L. Lyon, M.D., M.P.H.
Purpose of this Paper
I collect and shoot the Model 1903 Springfield. Since I purchased my first Springfield in 1992, a chrome plated beauty made in 1930 and obviously used, but not abused, by a color guard, I've heard of the low numbered Springfield receivers and the terrible danger they pose to a shooter. (Low numbered receiver are those with serial numbers below 800,000 made at Springfield Armory, and below 286,506 made at Rock Island Arsenal.) Some have stated emphatically no rifle with a low numbered receiver should ever be fired under any circumstance because of the risk of serious injury or death, but that high numbered receivers are perfectly safe.
My training is in medicine and medical research and I specialize in epidemiology, a discipline that looks at why bad things (epidemics) happen to people. As such I have a considerable amount of training in statistics. Whenever I heard emphatic statements about the safety of something such as a low numbered Springfield receiver my training and natural inclination are to get the numbers and put them into perspective with other risks we face on a daily basis. This is why I wrote this paper. I have attempted to put the risk of Springfield receiver failures into prospective using simple statistics, thus permitting the interested reader to make his own decision about the safety of the Springfield rifle receiver.
History of the Problem
The U.S. Model 1903 rifle, commonly called the Springfield, was used by the U.S. Military between 1903 and 1945. When the United States declared war on Germany in April 1917 there was a marked increase in the use of this rifle for training. Between July and December 1917 eleven rifle receivers shattered, causing one severe and 10 minor injuries to the soldiers using the rifle. Despite the intense demand for rifles caused by our entry into the war, production at both Springfield Armory and Rock Island Arsenal was halted in early 1918, and an investigation launched to determine the cause of the problem.
It was determined that the workers responsible for heat treating the receivers had used an "eyeball" method that relied on the color of the heated metal to determine if the steel had been heated to the correct temperature. Unfortunately, according to General Hatcher, the officer in charge of the investigation, "... it was quickly found that the right heat as judged by the skillful eye of the old timers was up to 300 degrees hotter on a bright sunny day than it was on a dark cloudy one" (See Hatcher, Julian Hatchers Notebook , Third Edition, Stackpole Books, Harrisburg, Pennsylvania, 1966, page 215). Heating to the higher temperatures led to crystallization of trace elements within the steel, making it too hard, and rather than deforming under high pressure, the receiver shattered, often permitting the bolt to exit the receiver, causing injury to the shooter. Between 1917 and 1929 three soldiers lost an eye to receiver failure, and six more had unspecified injuries consider serious. An additional 34 soldiers received minor injuries from receiver failures. There were no deaths reported from the failure of a Springfield receiver.
The heat treating method was immediately changed to a double heat treatment, and pyrometers were used to determine the temperature of the heated receivers. The change in heat treating was instituted between serial number 750,00 and 800,000 at Springfield and by serial number 285,506 at Rock Island Arsenal. Rifles manufactured after these serial numbers are referred to as "high numbered" receivers and are commonly stated to be safe to shoot.
A second problem that Hatcher found was the hardness of the brass cartridge cases, and the design of the Springfield chamber-bolt interface. He states:
"One thing made evident by these tests is the fact that the weakest feature of the modern military actions is the cartridge case itself. In the Springfield rifle the head of the cartridge cases projects out of the rear end of the chamber a distance of from 0.147 to 0.1485; in other words, there is a space of well over an eighth of an inch where the pressure is held in only by the brass." (See Hatcher p 205.)
During the 1920's officials within the Ordnance Department investigated the problem more thoroughly, including destructive testing of receivers. Three rifles with low serial numbers were fired with cartridges that produced known levels of pressure starting at 70,000 pounds per square inch. One receiver failed at 80,000 pounds and the other two at 100,000 pounds. All of these receivers withstood pressures well above that experienced with military ammunition, yet none failed until pressure was raised between 50% to 100% above normal operating pressure. In 1926 24 high numbered receivers were subjected to pressures up to 125,000 pounds per square inch. None failed. (See Hatcher pp 227-229).
On December 2, 1927 a board was convened by the U.S. Army to look into the problem, and determine how to identify the brittle receivers and determine if they could be strengthen by re-heat treatment. The board made the determination of where the problem had occurred in receivers, and its from their deliberations that we use the 800,000 serial number for Springfields, and 286,506 for Rock Island receivers. They also concluded it was not feasible to re-heat the "low numbered receivers", and that they should be withdrawn from service.
To discard approximately 1,000, 000 receivers would create a political problem of major proportions for the U.S. Military, especially at time when military was funded at an extremely low level. The decision also has be questioned from a numeric standpoint. There had been 58 reported receiver failures when the board made its decision. To suggest that 1,000,000 other receivers were defective because of the failure of 58 is extrapolating well beyond the available data. On February 7, 1928 after considering all the factors the Chief of Field Service, U.S. Army,, General Samuel Hof, made the following policy for the United States Army:
"Our ammunition is getting worse and accidents may be somewhat more frequent. On the other hand, some of these early rifles have been in use for many years and undoubtedly some of them have worn out several barrels. I do not think the occasion merits the withdrawal of the rifles of low numbers in the hands of troops until the rifle is otherwise unserviceable. On the other hand, I do not think we are justified in issuing such rifle from our establishments. I recommend that we instruct our Ordnance establishments to no longer issue rifles with these questionable receivers, that such rifles be set aside and considered as a war reserve and the question of the ultimate replacement of the receivers be deferred. When rifles are turned in from the troops for repair the receivers having these low numbers should be scrapped."
Hofs decision meant that low numbered receivers would not be issued, but that those already issued would remain in service. The Army was small enough that new troops could easily be issued high numbered rifles, but low numbered rifles already issued would remain in service.
The U.S. Marine Corp, because of an even more limited budget than the Army, did not follow this recommendation and never retired any of its low numbered receivers until they were replaced with the M1 rifle about 1942. The desperate need for rifles caused by World War II, saw many of the low number receiver rifles taken from war reserves and issued to U.S. and foreign troops. In 1942-44 the United States also equipped the Free French Army of Charles DeGaulle with low numbered Springfields.
The Director of Civilian Marksmanship (DCM) Program provided surplus military rifles to qualified civilians before and after the Second World War. During the 1960's the DCM offered to replace low numbered Springfield receivers with high numbered receivers. It is not known how many receivers were replaced.
Rate of Receiver Failures
Between 1917 and 1929 there were 68 burst receivers. Of the 68 no serial number were available for 11 receivers, four of those that failed in 1917. Two of the 68 were made at Springfield Armory and had serial numbers in the 950,000 range. Of the remaining 57 receivers 33 were manufactured by Springfield Armory and 24 by Rock Island. Hatcher provided the serial number and the date of failure for all 33 Springfield Armory receivers, and the same data for 22 of the 24 Rock Island receivers (see Hatcher, pp 442-447). This information was used in the analysis that follows. The overall failure rate by 1929 was 68/1,085, 506 or 6.3 per 100,000 receivers. The failure rate varied by site of manufacture, and each manufacturer is discussed separately.
Springfield Armory Receivers.
The overall failure rate of the 33 Springfield receivers was 4.13 failures per 100,000 receivers. This is shown in table 1 and figure 1. The failure rate was variable by year. Of the 15 years between 1903 and the end of 1917 when the heat treatment method was changed, there were no failures in five of the years (1908-10, 1912, 1915).
The highest rate of failures occurred among the receivers manufactured in 1904 (8.71/100,000), followed by 1911 (8.53/100,000), 1916 (7.53/100,000), then 1907 (7.26/100,000). The belief that the problem with brittle receivers was caused by inexperienced workers overheating the receivers in 1917 is not supported by the data. Only one of the 11 receivers that failed in 1917 was made in that year. The other ten were made before 1917, two in 1904. The distribution of these rates by year suggests that the problem of overheating the receivers was present during ten of the 15 years of manufacture, and was worse before 1917, especially in the earliest years of production, with 10 of the 33 known receivers being made before 1908.
The absence of receiver failures in some years suggests that the problem may have been specific to some workers who only worked during some years. At Springfield Armory the worst three years for receiver failure were 1904-1907 with 1905 being an exception. Receivers made in these four years account for nearly 45.5 % (15/33) of all the receivers that failed. The absence of failed receiver among those produced in 1908-10, 1912 and 1915 suggests that the problem was not caused by hastily trained war time workers unfamiliar with rifle manufacturing requirements.
Another measure of problems in manufactured objects is called time to failure. This is the length of time from manufacture till the product fails. It was possible to calculate a time to failure, expressed in years, using the serial number data on year of production, and the tables from Hatchers book (pp 442-447). Of the 33 Springfield receivers that failed, the time to failure in years ranged from one year to 22. The average time to failure for all 33 receivers was 12.48 years. Hatcher reports no receiver failures after 1929 suggesting there were no further receiver failures, (or the military no longer recorded the problem).
Because we lack data on the number of rounds fired by each rifle it is impossible to adjust the time to failure by actual number of rounds fired before failure. The demands of World War I undoubtedly increased the use of every available rifle, and rifles that had likely been fired once a year were now being fired weekly. The time to failure for the 1917 manufactured rifles is less than for those made in 1905 ( 16.4 years compared to 11 years). And while this might reflect poor heat treatment in 1917 it more likely reflects the much heavier use of the 1917 manufactured receiver compared to the 1905 receivers. It also suggests that the receiver failures were associated with the number of rounds fired.
I have already mentioned Hatchers observation that many of the receiver failure problems in 1917-1918 were due to brass cartridges cases that had not been hardened to the right degree. He provides no numeric data on which to judge this problem, but says it was recognized as a serious problem by the spring of 1917. Hatcher states that four of the receiver failures were due to accidentally firing an 8 mm Mauser round in the Springfield rifle. This causes pressures in excess of the 75,000 pounds per square inch proof pressure use to test the receivers.
Rock Island Receivers.
The overall failure rate of the 22 Rock Island manufactured receivers was 7.71/100,000, nearly double that of those manufactured by Springfield. (See table 2 and figure 1.) Rock Island produced rifles for 11 years, starting in 1905 and ending in 1914, and then during most of 1917 and early 1918, There were no receiver failures of rifles manufactured for five of those 11 years (1905-6, 1913-14, and 1917), a higher percent of years than Springfield Armory (33.3% compared to 45.5%). Receiver failures occurred in rifles made between 1907 to 1912, with the peak rate occurring in 1912 at 20.27 per 100,000, about two and half times the peak rate for any of the years of manufacture for the Springfield Armory rifles.
The average time to failure for Rock Island receivers was 11.6 years with a range from 5 to 23 years. While the range is narrower than that for Springfield receivers, the average years to failure were similar (12.48 years compared to 11.6 years). Of the 22 receiver failures in 1917-1918, 11 were made at Rock Island and seven at Springfield, and four were so badly damaged the manufacturer could not be identified. Rock Island receivers likely accounted for the majority of the receivers that failed during these two years.
Receiver Failures with Double Heat Treated Receivers
The failure of 11 receivers in 1917 was believed to be due to human error in the heat treatment process of the receivers, but after the change to double heat treatment there were four receiver failures, three Springfield manufactured receivers, and one Rock Island manufactured receiver (Hatcher does not provide the serial number). All four receivers were definitely double heat treated. In no cases did the receiver shatter as was the case with the low numbered receivers, but the failed receivers did bend.
The failure rate for the double heat treated receivers up to 1929 was slightly less than 1/100,000, for Springfield manufactured receivers, and 0.5/100,000 Rock Island receivers. The double heat treated receivers did fail, but at a much lower rate than the earlier receiver, and did not shatter, and so had less potential to seriously injury the shooter. Those who state that the double heat treatment method solve the problem generally ignore this evidence
I am aware of one receiver failure of a high number receiver about 1987-88 in Salt Lake City, Utah. The rifle was made by Springfield Armory and the serial number was over 1,000,000. The ammunition was said to World War II military ball ammunition. A piece of the receiver was blown off and there was evidence of crystallization along the fracture line. The stock and magazine were wrecked. The shooter sustained minor injuries, and sued the seller. The seller of the rifle found evidence the rifle had been fired with the bore full of grease. The seller's insurance company settled out of court.
Expected Failures after 1929
I also determined the distribution of failures by year from 1917 to 1929. Since the failure rate of receivers is a rare event, we assume that a receiver failure follows a Poisson distribution, and that the standard deviation is identical to the mean number of failures in a year. The number of failures by year for each manufacturer is shown in figure 2, and figure 3 and the combined rate in figure 4. Springfield Armory receivers had their highest failure rates in 1917 (5), and again in 1929 (5). The range of receiver failures per year varied from zero to five with no failures in 1919 and 1922 with an average of 2.64 failures per year.
The failure rate per year for Rock Island receivers varied from zero (1919, 1924, 1927-28) to seven in 1918. The average failure rate per year was 1.69.
Hatcher reports no receiver failures after 1929, but if the rates experienced between 1917-1929 continued up to 1939 there would have about 43 additional receiver failures. Or if all the low numbered rifles were withdrawn from service and replaced by high numbered rifles we would have expected up to 12 receiver failures through 1939. This provides a range of expected failures for this time period (12 to 43). An unknown number of low numbered rifles were reworked and put into service during World War II. There are no reports of receiver failures with these rifles.
The lack of receiver failures after 1929 may have occurred because the rifles with the most brittle metal had been eliminated in the 1917-1929 period. Another important factor is the exhaustion or retirement of soft brass cartridge cases manufactured during the crisis of World War I and still being used up to 1929..
Additional evidence for this explanation comes from the experience of the 1st Marine Division on Guadal Canal The Marine Corp made no effort to replace their low numbered Springfield rifles, and these rifles saw heavy use on Guadal Canal between August 1942 and February 1943. No receiver failures were reported in the training period before the battles, and during the four major battles that occurred in the seven month period in 1942-43. While it's not possible to estimate the exact number of rifles involved, up to 7,000 would have been in use by the three rifle regiments of the 1st Marine Division, Based on the failure rates of 1917-1918 between one and two rifle receivers would have been expected to fail.
Injuries Causes by Receiver Failures
Hatcher had data on the injury caused by the receiver failures for 43 of the 68 accidents. Three men lost an eye (7% of the total accidents) and 6 more (14%) had unspecified injuries considered serious or severe. The remaining 34 failures (79%) caused minor injury. The risk of serious injury from the failure of a low numbered Springfield receiver would be about 0.7 serious injuries per 100,000 rifles manufactured.
Putting Risk Into Perspective
It's hard for people to personalize risk to their own situation. The following are some risks of dying with common place activities that are of similar magnitude to serious injury from the failure of a Springfield receiver.
Risk of One Death per 100,000 population in a Single Year Caused By:
Riding a bicycle 100 miles
Smoking 14 cigarettes
Living 20 months with a smoker
Traveling 1500 miles by automobile
Traveling 10,000 miles by jet aircraft
The problem of Springfield receiver failures was a rare event throughout the service years of the Springfield rifle despite statements to the contrary. It was also concentrated in certain years of manufacture suggesting that an important component of the failure was human error in heat treatment. The heat treatment problems had been present long before the manufacturing pressures of 1917. The receiver failures were also compounded by a design flaw in the support of the cartridge case head in the Springfield rifle, and this problem was exacerbated by uneven manufacturing of brass cartridge cases during 1917-18.
Eleven receiver failures in 1917 prompted an investigation and a change in the heart treatment of the receivers. The decision in 1928 to replace the low numbered receivers as rifles were returned to arsenal for repair was an effort to provide soldiers with a greater degree of safety. The board of officers recommended that the low numbered receivers all be withdrawn from service, but the general responsible for reviewing this decision did not concur with the board's decision, and left most low numbered receivers in service until replaced by the M1 Garand in the early 1940's. He took a calculated risk, and the risk paid off. There were no further receiver failures after 1929.
It also suggests that ammunition manufactured during World War I likely played a major role in receiver failures.
I've used the detailed information that Hatcher provides in his notebook, and supplemented this with information from Campell and Brophy, and Ferris book on the Rock Island Arsenal Model 1903's (The Rock Island 03. Published by C.S. Ferris, 1992). There are some minor problems in Hatchers book for example see the table on pages 446-47. He lists receiver by date of failure, and the list is consistent until 1923 when he lists three failures, then four in 1924, then four in 1923, then three more in 1924. I checked his dates against the detailed report of the failures (see pages 448-482) and concluded his dates were correct, but his sequence was wrong. I have grouped them by the reported year of failure in the table.
Hatcher reports 24 Rock Island Arsenal receiver failures but only provides serial numbers for 22 (see page 443). One Rock Island receiver, number 445,136 is said to have failed in 1918, but Rock Island did not reach this serial number until 1919, after double heat treating was instituted. There was obviously an error in reporting the serial number, or the date of failure.
There are also two Springfield receivers (numbers 946,508 and 951,718) included in the low numbered receiver table, and counted among the 68 said to have failed. These I used to estimate the rate of failure for high numbered receivers.
Brophy has an error in his table of serial numbers on page 445. His table gives the beginning serial number for Springfield Armory for 1913 as 531,521, but the beginning number for 1914 as 510,561. I chose to use the serial numbers provided by Campbell for 1913 to 1917.
I also included the early 1918 receivers manufactured at Springfield Armory in the 1917 tally. Since Rock Island Arsenal had not been manufacturing rifles since 1914, I place their 1917-1918 rifles in a separate category.
I made no effort for allocate the 11 receivers to either manufacturer, or calculate an overall rate. If the failures were all from one arsenal or the other, then it would change their relative positions. If the failures were distributed similarly to the current allocation, then rates of each manufacturer would rise, but their relative position would stay the same.
Brophy, W.S. The Springfield 1903 Rifle. Stackpole Books, Mechanicsville, Penn, 1985. See pages 425-427.
Cambell, C.S. The 03 Era. Collector Grade Publications, Ontario, Canada, 1994.
Ferris, C.S. The Rock Island 03. 1992.
Hatcher, J.S. Hatchers Notebook. Stackpole Books, Harrisburg, Penn, 1962. See pages 205, 215, 221-223, 227-229, and 442-482.