In 1993, John Rodgaard, Daniel Martinez, historian of the National Park Service's USS Arizona Memorial, and noted Pearl Harbor historian and journalist Burl Burlingame launched a quest to solve one of history's great mysteries.1 Following is the latest report to be published by the U.S. Naval Institute on continuing research into the Japanese midget submarine attack on Pearl Harbor.2
After determining to pursue the research to support the National Park Service, Autometric, Inc., directed research team member and Autometric employee Rodgaard to assemble a team of experienced image intelligence analysts and photogrammetric scientists to determine whether a combat photograph supported Burlingame's claim that a Japanese midget submarine was present in the harbor during the attack. Subsequent analyses definitely supported that claim. Further, the experts observed the submarine had indeed launched torpedoes on two U.S. battleships. Hosted by Martinez, the team presented its findings at a press conference at the Arizona Memorial on 6 December 1994.
In 1998, Peter Hsu, respected naval architect, engineer, and noted marine forensic engineer for Techmatics/Anteon Corporation, joined the Autometric team. Independently, he had analyzed underwater explosion phenomena, cavitation effects, and concussion wave propagation seen in the attack photograph, and his analysis substantiated Autometric's findings on the two torpedo launchings. Hsu's analysis reinforced the team's observations that one torpedo exploded against the West Virginia, and he quantified the team's observations about the ensuing effects on the submarine by concussion waves from torpedo detonations.
On 7 December 2003, the Discovery Channel aired a special on this subject and presented several sides to the story. In it, John De Virgilio, a long-time scholar of the Pearl Harbor attack, stated the quality of the Japanese combat photograph we used in our assessment was poor. We argue the contrary. Carroll Lucas, Hsu, and Rodgaard examined the photo De Virgilio supplied to National Geographic Magazine and agreed that our reproduction was of much higher quality.
Another detracting contention is that, even though a number of small craft were in the area, no eyewitnesses reported sighting a midget submarine. Figure 1 is the combat photograph taken by an Imperial Japanese Navy aircraft during the attack. The small craft in the vicinity of the object in question (foreground) appears to be a 50-foot utility boat. Note the size of that craft compared to the object under investigation. Even if one discounts the team's photogrammetric expertise in obtaining precise measurements (object size and distance between objects) from oblique photography, the object is comparable in overall size to the 50-foot utility boat. (Note that the entire 80-foot length of the midget submarine is not observed on the surface as it breached. We measured and calculated it to be 59.3 feet.)
Also note the utility boat appears to be canvas-covered. If this craft had been traveling from fleet landing with returning sailors, all of them, except the boat crew, would have been sitting below the boat's gunwale and under the canvas covering. Only a very narrow gap between the bottom of the canvas and the boat's hull would have permitted most sailors from seeing what was going on around them. As for the boat crew, a sailor at the bow (the bow hook), the boat's engineer, and the boat's coxswain would have been able to move about the craft freely. Of those three, only the coxswain would have had a 360° view above the canvas cover. If he were standing at the boat's helm, he would in all likelihood have been looking skyward. It is also possible the boat crew and passengers were in the bottom of the boat, hoping and praying they would not be strafed.
Contrary to some contentions, not all of the midget submarines' torpedoes have been accounted for. Of the ten carried by the five submarines, only eight are truly accounted for, and this is despite the well-documented account by the crew of the light cruiser USS St. Louis (CL-49). In the Discovery special, De Virgilio assumed that the fifth midget sub fired two torpedoes at the cruiser when she made her high-speed dash from Pearl Harbor.
This alternative theory is that the I-16tou, the first midget released just after midnight the morning of 7 December, arrived near the entrance of Pearl Harbor around 0130 (about a half-hour ahead of her sister, the I-22tou, which was later sunk inside the harbor by the USS Monaghan [DD-354]). On their arrival near the harbor's channel entrance, the crew of the I-16tou waited about eight hours outside the harbor to watch the St. Louis or other large ships steaming out of the harbor after the main air attack. Knowing their submarine was in the vanguard, the crew of the midget waited another dozen or more hours to radio the mother sub about a successful mission.
This theory must be placed in context with the over-arching mission of the midget force. Namely, the midget subs were ordered to enter the harbor and attack the ships inside the harbor.3 All midgets that have been accounted for followed their orders under arduous conditions and in the fabled tradition of the samurai.
Another alternative theory contends that the Japanese used oxygen-burning torpedoes that left no wake. This is based on one of the abiding myths of World War II at sea. First, any object propelled through a fluid medium leaves a wake. The fluid is disturbed further by the cavitation of the object's propeller, and the track is enhanced. In addition, this contradicts the argument about the experience on board the St. Louis. How could the crew witness the torpedoes if they left no wakes?
A water geyser created by an exploding weapon such as a torpedo is a complex structure. It consists of "phases," and the first can be observed in the combat photograph. A "first plume" can be seen rising between the masts of the USS West Virginia. The towering geysers seen in similar attack photographs displayed the characteristics of a shallow-water detonation. These were consistent with all of the torpedo attacks delivered on 7 December 1941.
The shock wave caused by the torpedo explosion was measured in milliseconds. While the midget was at periscope depth, the shock wave expanded in a radial fashion and passed over the submarine in less than 100 milliseconds. This time step is chosen as a cut-off for closure.
A bulk cavitation envelope was created at the moment the torpedo struck the West Virginia.4 This envelope was calculated mathematically, based on the size of the explosive charge, the water depth, and the correlation with live test data. From the cavitation effect, the midget submarine was forced in a vertical upward motion. This action, compounded by the change in trim after she fired her torpedoes, exposed almost the entire sub's contra-rotating propellers as the surface wave from previous air torpedo detonations passed over and into the rotating propellers, creating the three proportionally spaced splashes (rooster tails) seen in the combat photograph.
The Water Tank Test
Discovery Television commissioned the University of Michigan's Hydrodynamic Laboratory to test the claims made by our team. Although the university experiment was purported to call our findings into question, the results actually confirmed our analysis.
The objective of its contra-rotating propeller experiment was to demonstrate the propeller spray effect generated by oncoming waves washing over the hull, while the model was in a pitching motion. The model was attached to a suspension apparatus at its sail (conning tower), allowing the model to pitch at this reference center. But the Michigan team failed to compare the design of its test model to the design and the dynamic behavior of the Type A Class midget submarine after she fired her torpedoes. Once the torpedoes had been fired, the midget's hydrodynamic behavior changed, shifting her horizontal center of gravity.
The difference between the Michigan test and the actual hydrodynamic motion of the midget submarine in Pearl Harbor was that the midget already was pitching erratically, and this motion was compounded by the upward motion caused by the underwater explosive effects of her own torpedo detonating against the West Virginia. As for the submarine test model, its center of gravity remained constant, fixed longitudinally one-third of the way from the bow of the model.
In our own analysis of a videotape of the experiment, we measured the splashes produced from the submarine scale model. Estimating the fluid effects created, then projecting them to the diameters of a full-size set of contra-rotating propellers of a Type A midget submarine (1.37 meters/1.24 meters), we estimated the height of the rooster tail created in the tank to be comparable to 37.5 feet. In essence, the tank test study substantiated the height of the largest rooster tail seen in the combat photograph we measured as being roughly 50 feet high.
The University of Michigan team concluded that the image of the splash in the combat photograph was the result of the torpedo's interacting with the water surface from the drop. This conclusion was based on the experiment's having a model torpedo hand-dropped from a moving platform suspended over the test tank. High-speed photography was recorded and analyzed. Figure 2 is a side-by-side comparison between images of the hand-dropped model torpedo and the combat photograph. The splash from the Michigan test is vastly different from the splashes (rooster tails) in the combat photograph. Our team agrees with Michigan's statement about the complexity of the effects of a solid object interacting with the fluid medium of water, in particular an air-dropped torpedo. In Figure 3, the difference is clear between the rooster tails in the combat photograph and the splashes created by an air-dropped torpedo.
Our team has compared many images of actual air-dropped torpedoes breaking the sea surface from various angles. In fact, the examples in Figure 2 were taken from the video segment of the Discovery television special, and our observations contradict those of the Michigan test. None of the photographed images of a torpedo entering the water from an aircraft resembles the Michigan lab test photography. And that is the case with the splashes seen in the combat photo.
In contrast, Figure 4 shows a typical attacking profile of an aircraft dropping a torpedo similar to that of the Japanese Type 91. Figure 1 shows a series of geometric splashes. In comparison, the Michigan team found that the splash geometry produced from their "hand-dropped" lab test was non-specific and did not match that seen in the combat photo. The conclusion was that that a torpedo entering the water from an aircraft caused the splash pattern. There was no similarity between the three splashes (rooster tails) in the combat photo and that of the Michigan lab test.
Comparing the images, one can see the relationship between the torpedo entering the water and the proximity of the aircraft dropping the torpedo. If the splashes seen in the combat photograph were caused by an air-dropped torpedo, as the University of Michigan claims, then the Japanese Kate torpedo bomber would be visible in the photograph. And it would be only a few yards ahead of the torpedo splash. The laws of physics and flight kinematics support this assessment.
Our team obtained a copy of a now-declassified U.S. Navy publication compiled by the Navy Technical Mission to Japan within eight months after the end of the war. It details the Japanese prewar torpedo research and development programs.5 During the 1930s and 1940s the Imperial Japanese Navy sponsored major research in the use of aerial torpedoes. It can be argued that Japan was ahead of its western counterparts in understanding the physical dynamics of these weapons. The Aircraft Technical Division of the Naval Ministry controlled their design and research specifications. Sea acceptance tests were conducted on the Sharu Shima test range under strict monitoring conditions, with small barges anchored at intervals. They measured speed by timing the first bubble to appear on the surface at the starting point and at each test barge along the test path. Hydrophones were not used. Speed was registered and measured between the difference of the velocity and the static head. The test also employed aerial photography to examine the running characteristics of the torpedo after it entered the water. From the air, the Japanese Type 91 torpedo exhibited a slight wake, but from the floating barges the wakes were scarcely visible.
The Type 91 torpedo, with four tail fins, took only 380 yards to assume its proper set depth level from the entry point into the water. The Japanese had added a small antiroll flipper (fin) on each side. Wooden fin attachments were used to stabilize the torpedo as it traveled from the aircraft and through the air to the water. The fins also acted as depth-control devices to prevent the torpedo from diving too deep. The Japanese aerial torpedo was not designed to meet the requirements of a specific aircraft. Instead, the aircraft was designed to suit the torpedo and its variants.
The Japanese calculated the optimum length-to-diameter ratio for an aerial torpedo was 11 or 12 to 1.6 As indicated by the translation of the declassified documents, the desired entry angle of a modified Type 91 torpedo into the water was between 17° and 20°, as determined by the optimum air speed and altitude of the aircraft delivering the torpedo. All in all, the Imperial Japanese Navy applied meticulous attention to the hydrodynamics of its torpedoes and their fluid interaction characteristics.
Further Analysis of the Attack Photograph
The I-16tou fired both of her torpedoes, one against the West Virginia, and one against the USS Oklahoma. Our team measured the alignment of the two torpedo tracks (with the parallax depression angle considered) and found them to be in agreement with the launch position of the submarine at periscope depth (just behind the three rooster tails).
As the midget submarine increased speed to escape, she ran into oncoming radial surface waves created by earlier (air-launched) torpedo detonations. The action of the submarine's contra-rotating propellers as each surface wave passed over the submarine produced three distinct splashes (rooster tails).
The surrounding surface disturbances at the midget's hull (as seen in the combat photograph) were re-created using a one-third-scale model against waves at the same scale during a test by Termite Films (producer of the Discovery Channel documentary) on Lake Ascade, California, in 2003. The subsequent images show these same types of disturbances surrounding the model.
The fluid interaction surrounding the midget submarine's hull magnified the pitching motion. The white water pattern surrounding the midget in the combat photograph is the same as the images taken during the Discovery-sponsored test (see Figure 2 for an enlarged image of the midget). As for the I-16tou and her one-third-scale replica, the oncoming waves washed over their hulls, and the disturbances seen are those resembling a figure-eight pattern (see Figure 5).
Correlation of Sources
While information concerning the fifth Japanese midget submarine's attack on Battleship Row was gleaned through photographic interpretation and photogrammetric and forensic techniques, the correlation of signal communications data lends credence to our team's original technical analysis. Human sources also could be cited and correlated, but the focus here is on the signal data.7
Thus, with four midgets recovered or found, what still remains is to explain more definitively the operations of the fifth midget, the I-16tou, in a way that confirms to image analysts our analysis was correct.8 We accomplished this by using the following translation from a Japanese military document, describing communications between the I-16tou and her mother submarine, the I-16:
The Type "Ih" No. 16 Submarine certainly communicated with its special midget sub from 17:06 (21:36) to 19:17 on the Eighth.9 It received the telegram of "Tora, Tora" (I succeeded in a sneak attack!), at 18:11 (22:41). After it received a weak signal with the specific code of "impossible to navigate" at 20:21 (00:51), there was no further contact at all.10
Over the past ten years we have received both challenges and support to our findings. As for the challenges, they have ranged from the historical to the technical. We will continue to maintain an open dialogue until the mystery is resolved completely.
1. In 1991, Burlingame presented a paper at a conference commemorating the 50th anniversary of the Pearl Harbor attack, indicating a midget submarine was probably in the harbor but further analysis of the photograph was needed. He later wrote the book, Advance Force Pearl Harbor (Annapolis, MD: Naval Institute Press, 2002).
2. See "Pearl Harbor: Attack from Below," Naval History, December 1999, and "Attack from Below," U.S. Naval Institute Proceedings, December 2000. b
3. The mission of attacking U.S. ships departing from Pearl Harbor was assigned to a ring of large Japanese fleet submarines outside the harbor and positioned around the Island of Oahu. See Burlingame, Advance Force Pearl Harbor, pp. 224-28.
4. See Frederick Costanza and John D. Gordon, "A Procedure to Calculate the Axisymmetric Bulk Cavitation Boundaries and Closure Parameters," David Taylor Research Center, Underwater Explosion Research Division; "Fluid-Structure Dynamics and Dynamic Analysis," Shock and Vibration Bulletin #53 (Washington, DC: Naval Research Laboratory, 1983); and Prof. Young Shin, "Overview of Underwater Shock and DDAM," Department of Mechanical Engineering, Naval Postgraduate School, Monterey, California.
5. Target Report—Japanese Torpedoes and Tubes, Article 1, U.S. Naval Technical Mission to Japan, 8 April 1946. The document was declassified on 3 May 1972.
6. U.S. aerial torpedo length-to-diameter ratio was less than the Japanese, and the Japanese considered U.S. aerial launched torpedoes as inferior.
7. See Ken Hackler, "Myths Surrounding the Midget Subs," Pearl Harbor Commemorative Issue, World War Two Publications Magazine. The author cites but dismisses the sighting of two submarine conning towers inside the harbor by the USS Breese (DD-122) about 0830, 7 December 1941, just before one of them (the I-22tou) was sunk by the USS Monaghan (DD-354). Then, some crewmen on the Monaghan thought they saw another conning tower but the author states, "It was a black cage buoy." Later at 0930, a Breese lookout also reported a periscope near the coaling yards and in the main channel leading toward the open Pacific Ocean. The author (because no one else saw it) states, "Clearly, what the lookout actually saw was just another piece of driftwood or debris."
8. See Burl Burlingame, "Midget Sub Found at Pearl," U.S. Naval Institute Proceedings, October 2002.
9. The times in parentheses are local Hawaiian times on 7 December 1941.
10. "Operations of the Submarine Corps at the Outbreak of War; The Attack of the Special Attack Unit; Opening Warfare Strategy for Submarine Division," Chapter 12, Japanese Self Defense Force, Institute of Defense, Department of War History, Hawaiian Campaign War History Series. Yoko Spalding, Japanese-English Quick Call Decision Interpretation, Inc., Springfield, Virginia, made translations of this document, which is available through the U.S. Naval Institute.
We thank the many individuals who have expressed opinions of our work in letters, e-mails, interviews, news articles, and television presentations. Special thanks go to the production company Creative Differences (formerly Termite Films) and its president, Erik Nelson, and producer, Richard Russell. Together, they highlighted our efforts in the Discovery Channel special discussed here.