A Special Report: What Really Sank the Maine?

Edited by Thomas B. Allen

On 25 January 1898, the battleship steamed into Havana Harbor. McKinley, trying to still the war drums, wanted the Maine to show the flag, prove that U.S. warships had the right to enter Havana, and then get out. On 15 February the Maine was to head for New Orleans in time for Mardi Gras. By then, McKinley hoped, anti-Spain fervor should have died down.

But at 2140 on the night of 15 February, a massive explosion tore through the ship, killing 250 men and two officers. (Mortal injuries raised the final toll to 266.)

A Court of Inquiry questioned survivors—including commanding officer Captain Charles D. Sigsbee—and interpreted the reports of divers. The theory that a mine had destroyed the ship stemmed primarily from eyewitness testimony. The report of diver W. H. F. Schluter was particularly significant. He said he could see green paint on a bottom plate that was "all torn ragged and it looked to be inward." Bottom plates on the outside were painted with antifouling green paint. So this produced the image of a plate being blasted from the outside and turned inward.

"You are sure they were not bent out?" the court asked Schluter.

"Yes, sir; I am sure," he replied.

"And the green paint you saw was on the part bent inward?"

"The green paint was on the part bent inboard. . . . My opinion is, I believe that she was blown up from the outside and in, because there was no explosion from the inside [that] could make a hole like that, from the way them plates stood around in different directions." The Court concluded that the extensive damage "could have been produced only by the explosion of a mine." But it was "unable to obtain evidence fixing the responsibility . . . upon any person or persons." After the court's finding was revealed in March, McKinley no longer could ignore the call for war. "Remember the Maine and the hell with Spain" became a rallying cry.

But was it a mine?

The question lingered until 1911, after the U.S. Corps of Engineers, in an unprecedented feat, built a cofferdam around the ship, pumped out the water, and exposed the wreckage. A Board of Inquiry based much of its analysis on photographs of physical evidence that the previous investigation had sensed but not seen: bottom plates that were bent inward, presumably by an external force, such as a mine. The board focused on a section of outside plating that "was displaced inward and aft and crumpled in numerous folds."

Although the 1911 report placed the location of the explosion farther aft, the 1911 inquiry's conclusion agreed with that of 1898: "The board believes that the condition of the wreckage . . . can be accounted for by the action of gases of low explosives such as the black and brown powders with which the forward magazine were stored. The protective deck and hull of the ship formed a closed chamber in which the gases were generated and partially expanded before rupture."

The question disappeared. Historians writing after 1911 took for granted that someone—Spanish sympathizers, perhaps, or disgruntled guerrillas hoping to goad the United States into war—had set a mine that blew up the Maine.

After reading a newspaper story in 1974 about the sinking of the Maine, Admiral Hyman G. Rickover decided to reexamine the issue. He recruited historians, archivists, and two Navy experts on ship design: Robert S. Price, a research physicist at the Naval Surface Weapons Center at White Oak, Maryland, and Ib S. Hansen, assistant for design applications in the Structures Department at the David W. Taylor Naval Ship Research and Development Center at Cabin John, Maryland. Among Price's Navy projects had been an analysis of the wreckage of the nuclear-propelled submarine Scorpion (SSN-589), which was lost in May 1968.

The Hansen-Price analysis, as Rickover called it, was the heart of a short book published in 1976. The 23-page analysis reached this conclusion: "We found no technical evidence . . . that an external explosion initiated the destruction of the Maine. The available evidence is consistent with an internal explosion alone. We therefore conclude that an internal source was the cause of the explosion. The most likely source was heat from a fire in a coal bunker adjacent to the 6-inch reserve magazine. However, since there is no way of proving this, other internal causes cannot be eliminated as possibilities."

Again, historians rallied around the Rickover solution, and after 1976 most discussions of the Spanish-American War concluded that there was no mine.

As the 100th anniversary of the sinking of the Maine approached, David W. Wooddell, senior researcher on the editorial planning council of National Geographic magazine, suggested that the magazine commission an analysis of the disaster based on computer modeling not available to Rickover and his team. Advanced Marine Enterprises (AME), a marine engineering firm often used by the U.S. Navy, accepted the mission.

The AME analysis, which was announced in the February 1998 issue of National Geographic, examined both the mine and the coal bunker theories. The report declared that "it appears more probable, than was previously concluded, that a mine caused the inward bent bottom structure and detonation of the magazines."

Some experts, including Rickover's researcher Hansen and respected analysts in AME itself, do not accept the conclusions of the AME report. Following are excerpts, published in cooperation with National Geographic, to give readers a chance to judge for themselves.

Carrying large quantities of coal on board was a source of constant hazards for ships of the time, and even today. Coaling operations inevitably left a ship covered with a layer of fine, black dust, creating a fire hazard and the potential for a coal dust explosion. However, a thorough cleaning would significantly reduce the risk of sufficient coal dust collecting on deck. Fires inside the coal bunkers were recognized as a constant hazard. Coal will naturally oxidize when exposed to air, producing heat. If this heat is not dissipated, it will feed the reaction, causing it to accelerate. Typically, the rate of the reaction will double for every 15°-20°F increase in temperature. As the temperature reaches 750°-800°F, incipient combustion occurs, followed ultimately by self ignition and flame. A higher moisture content in the coal will increase this heating tendency. Once one of these fires is started, it can be difficult to extinguish, often requiring emptying of the bunker to ensure removal of hot spots. The Titanic had a fire burning in one of her coal bunkers when she left Belfast for Southampton, prior to her fateful voyage. Stokers worked for several days to extinguish the blaze, and plans had been made to have the New York City Fire Department meet the ship upon its arrival in New York.

There are various ways of storing coal while minimizing the risk of spontaneous combustion. One method, widely used in modern coal-burning facilities, is simply to use the coal before any appreciable heating can occur. Another is to compact the coal when it is stacked, thereby minimizing the amount of air which can flow through the stack, and depriving any potential fires of oxygen. If the coal cannot be compressed, then limiting the size of the stack can help allow sufficient air flow to dissipate the heat. Stowing different sizes of coal together should be avoided. The finer pieces, which are more prone to react due to their increased surface area relative to their weight, will tend to collect near the bottom of the stack, while the larger lumps allow air to circulate through the pile.

  • All heat transfer occurs through the steel, since steel is much more conductive than coal.
  • Only the plating of the decks, frames and bulkheads was effective in transferring heat. Plating stiffeners had an insignificant effect on the thermodynamics and were therefore not modeled.
  • Heat transfer to spaces above the protective deck was considered negligible, due to the layer of wood laid down upon the deck, which would have insulated the spaces above from the deck below.
  • The structure on starboard side, aft of Frame 30, and forward of Frame 24 has minimal impact on the analysis as it is relatively far from the heat source.
  • The shell plate acts as a heat sink, being in contact with the water. The shell plate was not included in the model. Instead, boundary conditions were placed at the edges of the floors and girders, where they attach to the shell plate, and were set to a constant temperature of 80°.
  • Heat transfer by convection was negligible, because the vent was so small and no forced ventilation existed.
  • The initial temperature for a majority of the structure is assumed to be 80°. The initial temperature within the 6-inch Reserve Magazine and within the Coal Bunker is assumed to be 100°. This is considered to be conservative. A higher initial temperature would permit the temperature within the magazine to reach a critical value sooner.
  • The temperature input (load) is assumed to be a constant temperature of 1000°F. This is considered to be a low (conservative) value for the assumed temperature of the bulkhead close to a coal fire. A higher input temperature would result in temperatures within the magazine reaching a critical value sooner.
  • The 6-inch Reserve Magazine contains black powder, it is properly stowed and is therefore within 4 inches of the bulkhead.


Conditions on the Maine were far from ideal for preventing spontaneous combustion. The coal bunkers had last been filled with bituminous coal in November [1897], and the ship had spent most of the last three months at anchorage in Key West, Florida. The tall, narrow shape of the coal bunkers would have made it difficult to compress the coal to eliminate air flow, however, it would easily allow fine particles to collect at the bottom of the bunker, where hot spots would be most difficult to detect. With no means of forced ventilation, the small vent for the bunker could not have allowed sufficient excess air flow to dissipate additional heat, and the omnipresent moisture of the tropics would have ensured that the coal was moist. The forwardmost bunkers, immediately outboard of the 6-inch Reserve Magazine on the port side and the 10-inch Handling Room on the starboard side, were among the last to be emptied, so they were full, with over 37 tons of coal in the port bunker and over 24 tons in the starboard bunker.

To examine the heat-transfer theory, the AME team made a computer model to evaluate "the effects of a coal fire in the bunker immediately outboard of the 6-inch Reserve Magazine." The heat-transfer study was based on the following assumptions:

Conclusions from Heat Transfer Study

  1. Damage from a small mine of 75 to 100 pounds, black powder, exploding beneath the ship, specifically addressing whether it could touch off an internal magazine explosion in the 6-inch Reserve Magazine, giving results consistent with the known damage to the USS Maine . (The 100-pound weight was chosen, and the analysis also included gun cotton as an explosive material.)
  2. Damage from a larger mine explosion of more than 100 pounds of black powder, exploding beneath the ship, specifically addressing whether it could touch off an internal magazine explosion in the 6-inch Reserve Magazine, giving results consistent with the known damage to the USS Maine . (Two larger mine sizes were investigated, 200-pound and 500-pound. Both black powder and gun cotton were assumed as explosive materials.)
  1. Examine historical records of the damage evidence resulting from naval inquiries conducted in 1898 12 immediately after the sinking, and again in 1911. 13
  2. Gather reference materials on mines of the Civil War, Spanish-American War, and Russo-Japanese War (1904-1905) eras.
  3. Gather reference materials on explosive characteristics of black and brown gunpowder, and gun cotton (nitrocellulose).
  4. Gather reference materials on the plans of the ship, and thicknesses of hull plates and bulkhead steel, as well as, it was hoped, some metallurgical information on the types of steel used.
  5. Computer model enough of the ship, including internal bulkheads in the key areas around the coal bunker and forward magazines, so that modeling damage . . . can be performed to a reasonable level of confidence.
  6. Run the dynamic computer models under each of the scenarios, and reach conclusions on each of the scenarios.

back to top

The finite element analysis indicated that a coal fire burning close to the bulkhead could have raised the temperature of powder stowed in the 6-inch Reserve Magazine, close to the bulkhead, to the point of being hazardous in only a few hours. This is significant, since testimony [at the 1898 inquiry immediately after the sinking] indicated that routine checks had been made of the temperatures in the magazines, however, there were intervals of several hours between these routine checks.

The testimony provided at the first inquiry is sometimes contradictory as to contents of the 6-inch Reserve Magazine, however, all agree that some form of powder was stowed in this space, probably black powder for saluting. The testimony indicates that it was properly stowed, which suggests that it was stowed in copper canisters on the wooden racks. Any canister stowed in the rack against the bulkhead between the coal bunker and the magazine could have been heated to several hundred degrees, hot enough to cause the powder to explode. If any of these canisters were in direct contact with the steel bulkhead, they would have heated much more quickly.

The results of the finite element analysis cannot be considered conclusive, as there is no direct evidence supporting several assumptions included in the analysis. There is no evidence concerning exactly how and where powder was stowed in the 6-inch Magazine. There is no evidence that a coal fire was burning in the coal bunker. Even if there were a fire, there is no guarantee that it would have been close enough to the bulkhead to heat the magazine. Any such fire presumably would have started from a hot spot that previously had gone undetected. The heat buildup in the magazine could not have been in progress for more than a few hours prior to the last temperature check on the magazine. However, the analysis does indicate that if a hot spot went undetected in the bunker close to the magazine bulkhead, this hot spot could have developed into a fire. And if powder were stowed in the magazine, close to the same bulkhead, the powder would have been heated enough to explode. And all this could have occurred without the routine watch having noticed a temperature increase. While there is no direct evidence proving that a coal fire led to the explosion that destroyed the ship, the available evidence, combined with the results of the analysis, indicates that a coal fire could have been the first step in the Maine's destruction.

After the analysis of the heat-transfer, or coal-fire, cause for the destruction of the Maine, the report takes up the possibility that a mine could have been the cause. Using modern analytical techniques, the AME analysts conducted damage assessments based on:
The approach employed for the damage assessment was to:
The AME analysts chose three kinds of explosives for analysis: black powder, used for saluting guns; brown powder, used to fire projectiles; and gun cotton, used as a high explosive in mines and torpedoes.

While reference material made available to use had useful data on black powder and gun cotton, no technically meaningful data was found for brown powder, perhaps because manufacture of this material stopped sometime prior to World War I. In the absence of specific data, brown powder similar to black powder has been assumed because each is a mixture of potassium nitrate, charcoal, and sulfur, although brown powder has a lower sulfur contact (3% versus 12%) and a correspondingly higher potassium nitrate content. The reduced sulfur content may result in a slower rate of deflagration, but the higher potassium nitrate would support a more complete burning of the charcoal and thus release more energy.

The weapons-effects models used for the damage assessment require explosive characteristics inputs in terms of either TNT equivalencies or by providing the combustion and detonation energies. The available references provided no data on the needed energies but did provide sufficient information to derive TNT equivalencies.

The following table lists the TNT equivalencies of black powder and gun cotton, and the reference source.


1.02 TNT
Ballistic Mortar (Impulse) .50 TNT 1.25 TNT
Detonation Velocity 1,300 ft/s 23,950 ft/s
Heat of Explosion .60 TNT .95 TNT

Structural Steel Properties

Historical Perspective

Tensile and Yield Strengths


Relative Strengths of Welded and Riveted Connections

The sensitivity of the explosives to accidental detonation/deflagration is required only for the black/brown powder, since that is the only material known to be stored in the affected magazines. Examination of Drop Test (Impact Sensitivity) results yield an impact velocity sensitivity of 12.6 feet per second. The 10-second temperature for explosion test is 385°C (725°F). However, the longer the exposure to heat, the lower the ignition temperature. Since sulfur is the primary means of burning the charcoal in black powder and the autoignition temperature of sulfur is 232°C (450°F), we have assumed a long-term (greater than 10 seconds) heat sensitivity of 450°F.

Steel plating from the Maine's hull, thought to be available from the Naval Historical Center, Washington Navy Yard, could not be obtained. . . . Conclusions in this section have been drawn by analysis and deduction by an AME metallurgical engineer, based on re-view of technical references, consultation with forensic metallurgists, and Maine historians.

In the second half of the 19th century, steel ships (as opposed to wrought iron) were built in increasing numbers and by about 1890, the changeover was essentially complete. The battleship Maine structural drawings refer to the ship as an "armored steel cruiser" and the term "steel" appears on the drawings in reference to plating thickness. Steel used to construct the Maine is known to be of U.S. origin and fabrication.

Steel plate fabrication in the 1890s was accomplished by hot rolling operations. Typical compositions of "mild steel" were approximately 0.2% carbon, with strength characteristics comparable to today's designation SAE (Society of Automotive Engineers) 1020, with nominal yield strength (Sy) of 40,000 pounds per square inch (psi). . . .

AME's estimate for yield strength of Maine structural steel, in consultation with other experts, is Sy=40,000 psi; plus 10%, minus 20%. For our analysis, we have chosen the average of these upper and lower bounds, or Sy=38,000 psi. Similarly, our estimate for ultimate tensile strength (Su) is 60,000 psi. Both values have importance when using computational methods to characterize structural failures under explosive loading.

Toughness is a measure of a material's ability to absorb energy before fracture. It depends on both strength and ductility. A ductile material is one that can undergo considerable plastic deformation before fracture. Hot rolled mild steel plate in the normalized condition typically shows an elongation per unit length (strain) of 30% in tension test specimens, and is therefore quite ductile. . . .

Observation of Maine explosion damage in photographs shows very high distortion of plating and structural members prior to failure, indicating that the steel was a reasonably tough material. AME believes that 30% elongation is a good estimate of ductility and that the structure was in the range of or above the transition temperature, i.e., still "tough," in the warm Caribbean waters of Havana Harbor.

In order to use the internal blast damage model to analyze a ship of riveted construction rather than welded construction, a factor must be included to account for the reduced effectiveness of riveted construction when subjected to explosive blasts. . . . Finite element models were developed of typical lap and butt joints on board the Maine. The connections were evaluated for effectiveness under a membrane load in the plate. . . .

Evaluation Criteria

Analysis and Results

Shock Wave Holing of Hull

Internal Blast Damage Propagation

Effects of Residual Explosive Energy in the Inner Bottom

23 [T]he butt joints were single-strap, double-riveted joints. . . .

The interaction between rivets and plates is very complex. In order to model this structure, the rivets were assumed to be beam elements with end beams representing the rivet heads. The end restraints of the end beams connecting with the plates were manipulated to represent the bearing and bending that occur at riveted connections. . . . The material properties of the shell elements were represented using a stress strain curve with a yield strength of 34 ksi, or thousand pounds per square inch (corresponding to the type of steel used for the shell plating). The material properties of the beam elements were represented using a stress strain curve with an assumed yield strength of 30 ksi (corresponding to the type of steel used for the rivets). The beam elements representing the rivet head were input as elastic material with an elastic modulus of 30,000 ksi. . . .

In order to determine the ultimate capacity of the riveted connections, the pressure load on the connection was applied in increments until excessive deformation of the connection occurred. The . . . ultimate strength of the joint was determined by looking for significant areas of strain above 20%. The membrane load that caused this type of strain was considered to be the ultimate strength of the joint. . . .

The lap joint and the butt joint show similar ultimate capacities. The results indicate that the joint begins to deform inelastically at about 35% of the total load (34 ksi). At 45% of the load, significant portions of the plate around the rivet holes have predicted strains above the 20% strain criteria. At this strain, rupture may occur. This is the predicted ultimate capacity of the joint.

The finite element model shows that the failure of the joint occurs due to pull of the rivet through the plate material. Photographs of the damaged Maine support this predicted failure mode. The lap joint and the butt joint show similar ultimate capacities.

Using this approach, it was estimated that the ultimate capacity of a typical riveted joint on the Maine occurs at a membrane load of approximately 45% of the plate yield stress. This number was used to adjust the blast damage model failure criteria for riveted connections.

Shock wave holing of the hull was analyzed to determine the possible mine sizes and standoff distances that would hole the shell of the ship and transmit enough explosion energy into the ship to ignite the magazine.

The method used to calculate the conditions necessary for initiating a hole in the outer hull of a ship by explosion of a nearby mine relates (not equates) the incident explosion energy to the energy absorption capability of the hull plating in which the hole would be formed. Since part of the incident explosion energy will be reflected rather than absorbed by the plate, the incident explosion energy will be greater than the energy absorbed by the plate. By geometry, the incident energy per unit area of the plate is proportional to the energy of the charge divided by the square of the distance from the target plate, while the energy absorption capability of an area of plate is proportional to its strain energy (integral of stress times strain) up to rupture, multiplied by its thickness. Since we cannot equate these energies, we simply set up two dimensionless variables, one involving energies and one involving linear dimensions, thereby ensuring similitude. The relation between these two variables is acquired from numerous experiments. Calculation of the size of an opening follows similar logic. Calculations of the amount of energy that can be absorbed by the plate up to its rupture show this energy to be a small fraction of the incident energy (on the order of 5%). One can assume that the remainder of the incident energy is transmitted into the ship space beyond the opening as blast energy. This would seem to be an upper bound on energy transmitted, and test evidence supports this position.

The inputs needed for the shock wave hole and the blast energy vented through the hole calculations are: charge weight, standoff, plate thickness, yield strength, and plate backing. Two possible mine explosives were considered: black powder and gun cotton. . . . Three weights (100, 200, and 500 pounds) and three standoffs (contact, middepth [7 feet], and bottom [14 feet]) were analyzed for each explosive. . . . A thickness of .5 of an inch (taken from the ship structural drawings) and a yield strength of 38 ksi were used. The shell plate was air-backed.

The interior blast damage model was then run to determine if the residual energy for any of the mines that shock-holed the bottom plating was sufficient to fail the inner bottom plating and vent explosion energy into a magazine above. It can be assumed that once explosive energy and gases are vented into a magazine, the magazine will ignite. The modeling showed that even the smallest residual TNT (i.e., TNT equivalent weight of energy vented through hole . . . 6.5 pounds from the 100-pound black-powder contact mine) would cause failure of the inner bottom and, therefore, ignition of a magazine's contents. It can then be assumed that for the charge weights and standoffs shown in Table, if the mine holed the shell below a magazine and transmitted energy into the interior of the ship, it would ignite the magazine.

Damage assessments for the effects of explosions of various charge weights and initial source locations in the forward magazines of the Maine were conducted using an interior blast damage model . . . similar to those in use by the U.S. Navy as a diagnostic tool. . . . The model calculates the intense pressure that is generated during an explosion as a function of time and then systematically determines the effects of the quasi-static overpressure acting on the plating of the decks, bulkheads, and overheads that form the boundaries of compartments exposed to the blast, beginning with the compartment that contains the source explosive. An experimentally derived boundary plate failure criteria is used to determine when a pressurized compartment's boundary plates begin to fail and adjacent compartments are opened to the blast pressure, or when the blast pressure begins to vent either to the atmosphere or to the sea (in the case of a failure of the submerged portion of the hull plating). . . .

[T]he Maine was constructed using riveted connections. A finite element analysis of a riveted lap joint and a single strap butt joint typical of the types used on the Maine, showed that eccentricities in the loading produce high localized stresses in the plating around the head of the rivet that results in a connection failure when the membrane stresses in the plating are only about 45% of the yield stress for the material (the stress level at which permanent deformation begins to occur). This means that under blast loading, the individual plate segments that comprise a compartment boundary panel will tend to unzip along lines of riveted connections without any appreciable stretching, or bulging, of the material. . . .

The results of the shock wave holing study showed that the residual explosive energy in the inner bottom compartment under the 6-inch Reserve Magazine from a 100-pound black-powder mine exploded at maximum holing distance would be equivalent to the explosive energy developed from the detonation of 6.5 pounds of TNT. This equivalent explosive charge weight was used as an input to the model. The results of the assessment showed that this amount of explosive energy was sufficient to cause a failure in the inner bottom plating and to ignite the contents of the magazine. Since the 100-pound mine produces the least amount of residual explosive energy, it was concluded that any mine size and attack geometry that was capable of holing the outer shell would also penetrate the inner bottom and ignite the contents of the magazine.

A model for an initial explosion in the forward 6-inch Magazine produced a scenario showing that the explosion could not have occurred in the 10-inch Magazine. The report then looks at an initial explosion in the 6-inch Reserve Magazine.

This scenario assumes that the initial explosion occurred in the 6-inch Reserve Magazine, and that 7,200 pounds of black and brown powder (approximately 80% of the maximum allowed loadout) was exploded. It was also assumed that the inner bottom and outer bottom plating below the magazine was intact at the time of the explosion. This loading condition corresponds to the scenario postulating that the explosion was initiated by a local hot spot in the coal bunker outboard of the magazine. An initial explosion in the 6-inch Reserve Magazine of such a magnitude produces an initial damage propagation pattern that is supported by the historical evidence. Immediately following the explosion, the watertight bulkheads at Frames 24 and 30 and the longitudinal bulkhead separating the 6-inch Reserve Magazine from the 10-inch Shell Room fail. This allows the blast pressure to propagate forward into the Fixed Ammunition Room, aft into the Forward Fireroom, and across the ship from port to starboard. Failure of the starboard Shell Room bulkhead opens the 10-inch Magazine to the blast after just 0.73 ms (miliseconds). At this time, the blast also begins to propagate upward into the Electrical Storeroom on the Platform Deck level. The forward 6-inch Magazine is opened to the blast 1.25 ms after the explosion. The side shell forward of Frame 24 fails at 1.43 ms. Initial outer bottom failures occur on the port side between Frames 24 and 30 at 1.64 ms. Additional failures in the outer bottom plating occurs on the port side between Frames 21 and 24 approximately 2 ms later. The initial side shell failures on the starboard side occur just forward of Frame 24, but not until 5.02 ms after the explosion. The blast continues to propagate upward and forward in a general port to starboard direction. After 100 ms, pressures throughout the forward section of the ship have equalized. The outer bottom hull plating on the starboard is essentially unsupported but remains intact. The Main Deck has failed forward and aft of Frame 30, starting on the port side. . . .

A review of the damage propagation predicted by the model correlates very well with damage descriptions. . . with one notable exception. The 1911 damage evidence indicates that a section of outer hull B and C strakes was displaced upward, inward, and to aft and starboard. This type of damage cannot be explained by the model results. The model correctly identifies a failure of the outer hull plating in this area. However, the model can predict only if and when a plate begins to fail and the direction that the center of a plate is moving at the time of failure. . . . Where or how a particular plate fails is irrelevant to the analysis, and once it has been determined that a particular plate has failed, the geometry of the plate is no longer a concern. Only the total mass of the plate is used to estimate the rate at which venting into an opened compartment occurs. . . .

[A]t least 50% of maximum indicated loadout (10,600 pounds) would have to be exploded in order to produce outer bottom damage on the port side between Frames 21 and 30 and to initiate venting through the Main Deck above the Forward Fireroom. Further, if 100% of allowable contents are exploded, the model indicates outer bottom failures on the starboard side between Frames 24 and 30 and a probable failure of the keel web in this region. Since the predictions of starboard bottom failure are unsupported by the physical evidence, it is concluded that a loadout of between 60% and 80% of allowable is a reasonable estimate of the magazine contents at the time of the explosion.

To assess the effects of a submerged mine venting residual explosive energy into the magazine, it was decided to conduct an additional analysis of the 80% loadout case (7,200 pounds of explosive material) by including vent openings in the outer and inner bottom plating beneath the magazine. . . . The results . . . show that venting has little effect on the early stages of the blast propagation. . . . It appears that the explosive event is over before any appreciable venting through the bottom occurs. The extent of the final damage is comparable for both cases (i.e. bottom intact and bottom open to sea).

Next, the report looks at the effects of a sequenced explosion of the contents of all three forward magazines, based on the assumption that 80% of the contents of each magazine (a total of 38,800 pounds of black and brown powder) would be exploded. This scenario would apply to either an explosion caused by a mine or a coal bunker fire.

In general, the extent of the overall damage predicted by the model correlates well with the descriptions of the physical damage included in the 1911 Report. However, the prediction of outer hull plating failures on the starboard side between Frames 24 and 30 is contradicted by the physical evidence, which states that the outer hull plating on the starboard side is intact for the most part. This suggests that the estimate of 38,800 pounds of explosive content may be too high. A more precise estimate of the magnitude of the explosion cannot be made without the aid of additional assessments. It is also possible that the starboard outer hull plating could have unzipped along the keel strake but remained relatively intact. The model cannot predict how a particular plate fails.



We conclude that while a spontaneous combustion in a coal bunker can create ignition-level temperatures in adjacent magazines, this is not likely to have occurred on the Maine, because the bottom plating identified as Section 1 would have blown outward, not inward. Others have postulated that the inward bending of this panel could have resulted from the failure of the top of the watertight bulkhead at Frame 30 in an aft direction, pulling the inner bottom and outer shell plating and attached structure with it.

This scenario seems implausible, because while failures to a bulkhead from quasi-static overpressure have been known to pull pieces of connected deck plating with it in welded structures, on the Maine the bulkhead to deck rivets would fail long before sufficient bending moments could be imposed on the stiffened inner bottom/outer shell grillage.

Another theory [i.e., Rickover's] poses that the inrush of flooding water somehow could have not only reversed the downward bent plates and attached structural support to an inward position 180° from its original location, but to have done so without similarly affecting adjacent plating is also deemed implausible.

The most plausible explanation for the position of this plating is that non-shock-wave loading from an underwater mine, located beneath this plate section, caused the rivet connections to fail and pushed the plate section up into the ship; it in turn took the inner bottom plating above it up into the 6-inch Reserve Magazine at the inboard aft corner and ignited the contents. Most likely, the plate position just prior to ignition of the magazine contents did not move aft of the watertight bulkhead at Frame 30, but the ensuing magazine mass detonation pushed it and the watertight bulkhead aft and nearly horizontal in its observed final position. In this scenario, the smooth appearance of the outer shell compared to the severe crumpling of the inner bottom plating can be explained, due to the fact that the inner bottom plates were only 5/16 of an inch thick, while the outer bottom plating is a half-inch thick and nearly three times as strong in banding resistance. Also, the inner bottom plating is forced into a tighter bend radius than the outer plates.

Since the Shell Holing and Internal Blast Damage Models cannot simulate this type of failure, finite element modeling would have to be conducted to confirm our determination more conclusively. This level and type of modeling was neither planned nor included in our estimates for performing the damage assessment. . . .

This study strengthens the arguments in favor of a submerged mine as the cause of the sinking. . . .

The summary conclusion of this study is that the explosions that caused significant damage to the Maine, and were related to the ship's sinking, could have been by either of two possible causes:

   1. a magazine explosion induced by proximity to a coal bunker fire
   2. a magazine explosion induced by an under-ship mine.

The coal bunker could have experienced a local area of combustion, which would have gone undetected from the top of the bunker. If in proximity to the bulkhead that divided the bunker from the 6-inch Reserve Magazine, the magazine is predicted to have experienced a heat rise to a level sufficient to cause ignition of the gunpowder in the magazine. The temperature of the combustion in the bunker was assumed to be 1,000°F. After four hours, starting at ambient conditions of 100°F within the magazine, the temperature six inches from the bulkhead within the magazine was predicted to be 645°F. Gunpowder is predicted to ignite at a temperature of 450°F. The predicted high temperatures are sufficient to have caused the contents of the 6-inch Reserve Magazine to explode, which would have in turn caused the Forward 6-inch Magazine and 10-inch Magazine to explode. The damage resulting from this series of explosions is unlikely to result in the inward bent bottom structure.

Mr. Allen is a prolific writer and military historian and is the author of "Remember the Maine?" which appears in the February 1998 issue of National Geographic magazine. Mr. Allen and the U.S. Naval Institute gratefully acknowledge National Geographic—particularly the magazine's Senior Researcher, David W. Wooddell—for allowing Mr. Allen to adapt its Maine report for publication in Naval History.

  1. Norman Polmar and Thomas B. Allen, Rickover: Controversy and Genius (New York: Simon and Schuster, 1982).
  2. H. G. Rickover, How the Battleship Maine Was Destroyed (Washington, D.C.: Government Printing Office, 1976). A revised edition was published in 1995 by the Naval Institute Press, with a new foreword by Francis Duncan, Dana M. Wegner, Ib S. Hansen, and Robert S. Price. A new appendix gives details of World War II ship damage not available in 1976. The authors use this data to bolster their findings that a mine did not destroy the Maine.
  3. Handbook 1081, Primer On Spontaneous Heating And Pyrophoricity (U.S. Department Of Energy (DOE).
  4. Environment Safety and Health Bulletin EH-93 -4, The Fire Below: Spontaneous Combustion in Coal; U.S. Department Of Energy.
  5. Handbook 1081 (U.S. DOE).
  6. William H. Garzke Jr., David K. Brown, Arthur D. Sandiford, John Woodward, and Peter K. Hsu, "The Titanic And Lusitania: Final Forensic Analysis," Marine Technology, October 1996.
  7. Fire Protection Handbook, 16th edition (National Fire Protection Association [NFPA]).
  8. Ibid.
  9. Handbook 1081 (U.S. DOE).
  10. Fire Protection Handbook (NFPA).
  11. The Report of the Naval Court of Inquiry Upon the Destruction of the United States Battleship Maine in Havana Harbor February 15, 1898, Together With Testimony Taken Before the Court (Washington, D.C.: Government Printing Office, 1898-Library of Congress).
  12. The Report of the Naval Court of Inquiry, 1898.
  13. Report on the Wreck of the Maine, 14 December 1911.
  14. All figures in table from Cooper and Kurowski, Introduction to the Technology of Explosives (VCH Publishers, 1996) or Explosives and Demolitions (Department of the Army, FM 5-25, Feb. 1971).
  15. T. L. Davis, The Chemistry of Powder and Explosives (New York: J. Wiley & Sons, 1941).
  16. Cooper and Kurowski, Introduction to the Technology of Explosives.
  17. Sax and Lewis, Hazardous Chemicals Desk Reference (New York: Van Nostrand Reinhold, 1987).
  18. N. Cary, Head, Curator Branch, Naval Historical Center, Washington, D.C.
  19. S. Hering, M. Mat. Sci., B. Met. E., Advanced Marine Enterprises, Arlington, Virginia.
  20. D. A. Fisher, The Epic of Steel (New York: Harper & Row, 1963) Chapter 18.
  21. D. Wegner, Curator of Models, Carderock Division, Naval Surface Warfare Center, Bethesda, Maryland.
  22. H. Keith, Ph.D., Forensic Metallurgist, Marathon, Florida; T. Foecke, Materials Scientist, National Institute of Standards and Technology, Gaithersburg, Maryland.
  23. The Report of the Naval Court of Inquiry; Rickover.
  24. Robert H. Cole, Underwater Explosions, (Princeton, NJ: Princeton University Press, 1948).
  25. Rickover analysis.


Thomas B. Allen is well known for his writing about the intersection of espionage and military history, including George Washington, Spymaster.

More by this Author

Ship of Gold
Ship of Gold is an exciting, fast-paced tale of action, intrigue, and adventure on the high seas,... Read More

Events and Conferences

26 October - Guest Lecturer
9:00AM-11:00AM | SNAME Annual Conference at the Hyatt Regency Houston | Houston, TX 77002 Read More


Conferences and Events

Maritime Security Dialogue

Fri, 2018-10-05

Maritime Security DialogueNaval Aviation: Readiness Recovery for Combat A discussion with VADM DeWolfe Miller, USNCommander,...

The New China Challenge

View All

From the Press

22 September - Annual Symposium

Sat, 2018-09-22

22 September - Annual Symposium

Sat, 2018-09-22

John B. Lundstrom

Why Become a Member of the U.S. Naval Institute?

As an independent forum for over 140 years, the Naval Institute has been nurturing creative thinkers who responsibly raise their voices on matters relating to national defense.

Become a Member Renew Membership