ARTICLES – A NEW ESTIMATE OF U.S. TORPEDO SUCCESSES IN WORLD WAR II

John D. Alden is a retired U.S. Navy Commander and a veteran of three war patrols in the Pacific Theater of World War II. He has written extensively on naval history, including five books and numerous articles. He is known to several generations of submariners as the author of The Fleet Submarine in the U.S. Navy. He lives in Delmar, N. Y.

Our submarines’ major contribution to the destruction of the Japanese Navy and merchant marine is well known, as are the handicaps they suffered from defective torpedoes. Their successes are usually stated quantitatively in terms of the number and tonnage of enemy ships sunk or damaged. However, two data sources, one old and one new, now offer the means for calculating a different and more direct measure of the success rate of our submarine attacks during World War II. The old source is a master list of all U.S. submarine attacks issued by the Office of Strategic Planning, ComSubsPac (sic) and titled “U.S. Submarine Attacks, Listed by Date and Hour of Attack, Based on Task Force Commanders’ Assessments, Data from Records of SORG.”1 The new source is statistics derived from the data base for the recently published book United States and Allied Submarine Successes in the Pacific and Far East During World War Two by Craig R. McDonald and myself.

Source 1-The Submarine Operations Research Group (SORG)
Little specific information has been published about SORG and its work during the war. According to Admiral Charles A. Lockwood, he added the group to his staff organization in 1943.3 The results they produced, he wrote, “were startling at times and always highly valuable in shaping the trend of our efforts. Before the war ended they could tell us- and prove their statements what firing ranges produced the best results, what type of torpedo spread got the most hits, what agencies probably caused our heaviest losses- there seemed to be nothing which they could not reduce to a punch card on an IBM machine. The results of their studies were published monthly or oftener, in our Submarine Bulletins.” Considering their subject matter, these bulletins, and SORG’s output in general, would have been classified secret or higher and given limited distribution.

Although the SORG document that is the basis for this analysis is undated, it was obviously produced very shortly after the end of the war. Originally classified secret, the report was not publicly available until declassified in March 1972. It is a compilation of all reported U.S. submarine attacks during the war, whether by torpedo, gunfire, demolition, or other means. Its columns provide data on some 15 key elements of each attack such as the time and location of the attack, the presumed type and tonnage of the target, and the claimed result: sunk, damaged, or not hit. In the case of a torpedo attack, it gives the number and type of torpedoes fired and the hits claimed. I have checked it extensively against the actual patrol reports and can attest that it is an accurate compilation of data from those reports.

There is no indication in the document itself of its intended purpose, but in context it (or a similar compilation) appears to have been the basis for Admiral Lockwood’s post-war effort to publicize the Silent Service’s achievements, which until then had been largely cloaked under strict security. Summary totals for each month and year give the number and tonnage of ships claimed sunk and the same for those damaged. Adding up the yearly figures produces these grand totals: 2,662 ships of 10,751,700 tons sunk, and 1,005 ships of 5,733,800 tons damaged.

According to Clay Blair, “Lockwood and his staff tabulated the final results for all submarine commands and submitted the figures to the Navy Department. Lockwood claimed that U.S. submarines had sunk about 4,000 Japanese vessels for about 10 million tons.” However, the Joint Army-Navy Assessment Committee (JANAC) arrived at much smaller figures: 1,392 ships of 5,611, 117 tons sunk.5 “When confronted with the revised sinking figures according to JANAC, Lockwood laid most of the blame for the large discrepancy between claims and actuality on defective torpedoes. A total of 14,748 torpedoes had been fired. Had all these run, hit, and detonated as designed, the claims might well have been closer to the actuality, he maintained.

While Lockwood was understandably disappointed that his claims were so significantly discounted, he accepted the JANAC figures in place of his own, only adding: “Where JANAC does not credit a sinking as claimed by the Submarines Forces, it appears fairly safe to list that ship as having been damaged.”7 Lockwood was wrong on that count, but the record of his beloved Submarine Forces was indeed much better than credited by JANAC. Actually, the two sets of figures are not even directly comparable, because the ComSubPac/SORG totals counted vessels of all sizes, down to the smallest sampans, whereas JANAC arbitrarily excluded all merchant types of less than 500 gross tons, eliminated other valid sinkings as well, and made no attempt to assess damage.

The ComSubPac/SORG document (hereafter referred to simply as SORG) contains a wealth of information about U.S. submarines’ performance during World War II, but it has important drawbacks. From a working standpoint, the printed version is too voluminous and detailed to be readily analyzed by hand. Until very recently the data were not machine-readable, a deficiency that has now been overcome by my colleague, Craig McDonald.

The most serious weakness, however, is that SORG’s claimed results were based solely on uncorroborated information from the U.S. side. They did not reflect any of the wartime intelligence from Ultra intercepts of Japanese radio messages, which confirmed and identified many victims of our submarines but also revealed many instances where ships that submarine skippers believed to have sunk actually escaped undamaged. The extreme secrecy under which the Ultra findings were held precluded the release of such information, apparently even to so important and highly classified a group as SORG.

At this point, a word of explanation is necessary regarding the SORG definition of an attack. Submarine commanders generally did not fire torpedoes singly but preferably in salvos of two or more spread out to increase the chances of obtaining a hit. In the case of multiple targets such as convoys, a salvo could be split among several targets. If a target was not sunk immediately, additional torpedoes would be fired until the target sank or the attack had to be broken off. SORG therefore did not assess the result of an attack on a designated target until the final torpedo had been fired. However, it counted as a success any attack that resulted in sinking or damaging the target, no matter how minor the damage. Also, where SORG assigned partial credit to a submarine, it credited each such case as a success. Attacks as defined by SORG realistically reflect actual submarine performance, whereas the mere fact that 14,748 torpedoes were fired cannot lead to a meaningful assessment, because there is no way of determining which individual torpedoes hit and which ones did not. The results claimed by SORG, and probably used by Admiral Lockwood, are as follows:

SORG Claimed Results

*Includes 27 attacks by burning, demolition, etc.

The above figures show an apparent success rate of 95% for the gun attacks. This high rate is not surprising, given that the gun attacks were made on the surface mainly in daylight. The same cannot be said for the torpedo attacks. To arrive at a more realistic success rate, it is necessary to turn to the second data source.

Source 2-U.S. and Allied Submarine Successes
Our book is the fourth edition of a work originally published in 1989, to which each update added newly available data from declassified Ultra records and post-war Japanese sources. It was initially developed as a list of only those attacks that were claimed by the submarine commanders as successful, i.e., that resulted in the sinking of or damage to the target. These attacks were extracted from the SORG list and matched against the records of actual Japanese ship losses. Over the years we corrected many of SORG’s assessed results and picked up some cases from other sources, so that the numbers of attacks in our data base now vary slightly from SORG’s counts. Craig McDonald has just started the onerous task of fully merging the two sets of data, our book’s and SORG’s. This will ultimately enable us and other researchers to analyze the performance of the different Marks/Mods of torpedoes fired, examine how the success rate changed from month to month, and make many other detailed studies.

As a first cut, I have used the overall data to derive a new estimate of the success rate of U.S. submarine torpedo attacks. (Note that even negligible damage caused by a dud hit is counted as a success.) Our data indicate that approximately 116 cases that SORG had rated as misses were actually successful. On the other hand, 971 of SORG’s claimed successes turned out to be misses. Applying these adjustments to SORG’s claimed results yields the following revised figures for the torpedo attacks:

SORG Corrected Torpedo Results

Allowing for probably minor inaccuracies in the data, it appears that about 43% of U.S. torpedo attacks succeeded in hitting their targets while 57% missed.

It is now known that many misses were indeed caused by torpedo malfunctions. Some, such as premature detonations close to the target, certainly could have looked like direct hits and were so claimed by submarine skippers. Other failures, such as torpedoes that ran too deep or magnetic exploders that failed to function even though the torpedo passed directly under a ship, undoubtedly occurred but are impossible to prove. On the other hand, submariners tended to be overly optimistic in interpreting the results of their attacks.

A more detailed examination of our statistics shows that damage claims were considerably more likely to be incorrect than were claims of sinkings. This conclusion is supported by descriptions in scores of patrol reports where damage was claimed on very weak evidence: explosions heard but not seen, targets not in sight after surfacing later, disappearing radar pips, etc. The overall reliability of torpedo attack claims is indicated by these rounded-off statistics:

Of all claimed sinkings, 71 % were in fact sunk, 8% damaged, and 20% missed.
Of the damage claims, 15% actually were sunk, 20% damaged, and 65% missed.
Of the claimed misses, however, 3% turned out to be sunk and 4% damaged.

Given the ordinary limitations under which our submarines had to operate at different stages of the war-such as unfavorable ranges or target angles, torpedo shortages, enemy countermeasures, bad weather, etc.-compounded by the handicap of unreliable torpedoes-an overall torpedo attack success rate of 43% seems quite commendable.