One of a battleship designer's most serious challenges was protecting a ship against underwater attack, particularly by torpedoes.

Battleship designers were first forced to deal with damage at or just below the water line due to gun shells striking the side. A heavy armored belt extending along much of the ship's length above and below the water line accomplished this task. Later, when it was learned shells could travel an extended underwater trajectory and strike below the belt, a thinner internal lower belt or a thickened armored torpedo bulkhead was provided to stop this form of attack.

Defending a ship against detonation under the hull from magnetic torpedoes and mines, hand-placed limpet charges, and ground mines has never been adequately resolved. Blasts under the hull trap the expanding gas bubble under the hull, venting all of it into the ship. Double- and triple-bottoms ameliorate the effects somewhat, as can bottom armor on the inner surface of the double- or triple-bottom. Even if extensive flooding can be avoided, the massive shock effect often deranges the machinery plant and weapons mounts.

Blasts against the underwater side of the ship are another matter. Much of the expanding gas bubble of the explosion is vented upward, through the surface of the water as a plume of spray. Side striking weapons could be defeated by heavy armor such as that of the belt, but most were intentionally designed to strike below the depth of the belt. On those rare occasions when the belt was hit directly, the armor was typically displaced inward with much leakage into the voids or tanks behind it. The belt could not be extended down to the turn of the bilge due to its excessive weight. Thus another form of protection had to be provided against torpedoes, mines and near-miss bombs between the belt and the bilge.

A torpedo defense system (TDS) typically extended from the bottom of the belt to the turn of the bilge vertically, and from just forward of the foremost magazines to just aft of the aftermost magazines. Beyond this region, fore and aft, the ship became too narrow for installation of a TDS. No adequate means was ever devised to protect a ship's seaworthiness forward, or its rudders and screws aft, from a torpedo hit.

Because effective self-propelled torpedoes evolved roughly concurrent with the Dreadnought type of battleship, these were the first vessels to receive such protection. HMS Dreadnought herself had only partial protection against torpedoes. Pre-dreadnought ships seldom had anything that could truly be called a TDS.

Even early dreadnoughts, which universally burned coal, had very poor TDS systems. Often these consisted of little more than the extension of the double-bottom up the side to meet the lower edge of the belt. Inboard of this was a coal bunker, intended to absorb any of the torpedo blast that pierced the double bottom, with the inboard bulkhead of the bunker serving as the flooding boundary, or "holding bulkhead." This scheme was fatally flawed by the need to pierce the so-called holding bulkhead with coal scuttles to permit shoveling the solid fuel to the boilers. These were often blown open by blast overpressure, provided they were even closed or adequately watertight to begin with. Closing open scuttles against inrushing water was often academic.

More developed coal-burners incorporated some sort of expansion space between the double-bottom at the side and the coal bunker's outboard bulkhead. In this case the unpierced outboard bulkhead of the bunker served as a true holding bulkhead. Unfortunately the bulky character of coal fuel seldom permitted the expansion space outboard of the bunker to be large enough.

Some designers felt the presence of solid coal could help deplete the energy of a torpedo's explosion through the blast pulverizing the coal. However, this could not be relied upon. As more coal was consumed, there was less present to serve a protective function. Also, the dust of pulverized coal posed an explosive hazard.

Designers gradually came to realize that coal fuel hampered adequate torpedo protection, a fact which promoted coal's replacement with oil as much as did the liquid fuel's easier handling and greater thermal efficiency. Designers also determined a TDS needed to fulfill the following basic requirements:

A) It must absorb the overpressure of the gas bubble generated by the explosion.

B) It must arrest the fragments of the warhead and the ship's own structure formed by the explosion.

C) It must prevent the protected compartment inboard of the system from flooding.

Through the process of experimentation and experience, battleship designers learned the following:

A) The best means of depleting the gas overpressure bubble was unrestricted expansion into empty volume.

B) Liquid effectively depleted the kinetic energy of fragments, and disrupted the gas overpressure bubble through turbulence.

C) Armored bulkheads were required to prevent fragments from penetrating the system. Also, a properly elastic armored bulkhead acted as a tough membrane for depleting the gas overpressure bubble.

D) The side shell of the ship must be sufficiently elastic to tear under explosive force and form as few fragments as possible.

E) Inboard flooding was best prevented by placing an unpierced holding bulkhead as far as possible from the side shell. This proved the most important factor in the effectiveness of any TDS, regardless of its other design qualities.

The major limiting factor of any TDS, no matter how innovative or resilient was the size of warhead it was designed to handle. Just as larger and more powerful guns could fire shells through existing armor schemes, larger and more powerful torpedoes could penetrate existing TDS systems. Barring extensive--and expensive--reconstruction, nothing could be done to improve a TDS except to add an external bulge or "blister" to the hull to provide more stand-off distance between the blast and the holding bulkhead.

As oil-firing replaced coal-firing, dreadnoughts were typically fitted with some form of layered TDS, combining empty (void) compartments, liquid filled compartments, and at least one armored bulkhead. Often this armored bulkhead also served as the holding bulkhead, an arrangement carrying the inherent disadvantage of ensuring the protected space inboard would flood if fragments pierced the armor.

An obvious advantage of using liquid-loading--rapidly exploited--was the ability to store fuel in the liquid protective layer and then replace it with sea-water of similar density, thus retaining the system's protective qualities as fuel was expended. Less obvious was the proper sequencing of liquid and void layers, or their appropriate depth.

Placing a void layer outermost permitted unrestricted expansion of the gas overpressure bubble, but also allowed considerable disruption of the side shell and ship's spaces above the outer layer. It did nothing to decelerate fragments of the warhead and the ship's skin. Also, it resulted in sudden, significant initial listing moments by introducing liquid mass far off the centerline.

Placing a liquid layer outermost restricted the expansion of the gas overpressure bubble, and decelerated fragments, but transmitted more of the blast's force to the interior of the TDS. However, this scheme reduced the initial list due to the normal presence of a liquid load in that space.

As World War I progressed and ships began to experience underwater damage from mines and torpedoes, designers strove to create the most effective TDS possible. Across the Atlantic, where the US was still at peace, careful, conservative designers produced one of the structurally most sound torpedo defense systems ever devised, introducing it in the 1915 design for the Tennessee Class battleships.

The five-layer system took advantage of the reduced beam requirements of the turbo-electric drive system also planned for the Tennessee Class. This permitted the designers to give the system more all-important depth. The outermost and innermost layers were left void, with the three middle layers liquid-loaded. The novelty of the system lay in three thin, highly elastic armored bulkheads fitted between the four innermost layers.

Each bulkhead was carefully designed to provide maximum resistance to overpressure before tearing. The bulkheads were spaced so that once torn, a failed bulkhead would not impinge upon the next bulkhead inboard, permitting that structure in turn to provide maximum resistance, undamaged by the preceding structure. Similarly, the last armored bulkhead would not impinge on the holding bulkhead. The collective resistance of the three armored bulkheads and liquid layers stopped fragments before they could reach the unarmored holding bulkhead.

The system performed very well. The outer void space produced an initial sharp listing moment, but this was readily corrected by counterflooding corresponding outboard void spaces on the opposite side of the ship, a technique aptly demonstrated by the USS West Virginia (BB-48) at Pearl Harbor. The armored bulkheads performed as designed and the holding bulkhead remained intact when struck cleanly within the system by Japanese aerial torpedoes. USS California (BB-44) sank at Pearl Harbor due to her unprepared state; neither torpedo penetrated the TDS.

USS West Virginia sank due to the torpedoes striking her belt and punching it inward, causing flooding of the inboard compartments above the TDS on her third deck. Several torpedoes also opened the side shell above the belt, flooding the second deck, and one struck bodily above the belt. None of the torpedoes hit West Virginia's TDS cleanly and it may have been breached by virtue of the inward-driven belt buckling the third deck that sealed the top of the TDS, and weakening or tearing away the upper foundations of the torpedo and holding bulkheads. This out-of-parameters situation came about due to the ship's overloaded condition, scheduled to be corrected by blistering. In addition to the seven hits on the belt and one above it, one torpedo struck the rudder well outside the TDS.

Also in 1915, the British introduced an innovative TDS design in the Renown Class battle cruisers. The designers provided an integral bulge in the hull design below the waterline. This feature gained added stand-off distance for the TDS, and the upward venting of the gas overpressure bubble expended itself against the heavy armored side belt where it sloped outward above the bulge. One drawback of the design was a wasp-waisted cross-section that produced a smaller water-plane area that initially adversely effected stability, although flare higher on the hull began to take effect as the vessel became more deeply immersed..

A more questionable British innovation occurred in 1917, when HMS Ramillies of the Revenge Class, received external blisters containing "water excluding materials" in the form of closed metal tubes 8 and 9 inches in diameter, and wood pulp. The theory was that these materials would preserve buoyancy by preventing water from filling the entirety of the void. It was also hoped the torpedo would expend much of its energy crushing the tubes. In reality the wood pulp became waterlogged and rotten, eliminating its usefulness and the tubes appear to have been of no value. HMS Ramillies and HMS Resolution were both severely damaged by torpedoes in World War II, and HMS Royal Oak capsized from at least two hits (maybe three) in Scapa Flow.

The British Nelson Class of 1922 was the first Treaty-limited design, but used a conventional layered TDS. However, the Nelson's belt was placed inboard of the side shell, permitting torpedo blast to travel up the exterior of the armor yet still destroy the skin of the ship, possibly permitting flooding over the top of the TDS.

The Italians made the next, much more negative leap in 1934, with the Pugliese System introduced in the Vittorio Veneto Class and the reconstructions of the Conte di Cavour Class and Andrea Doria Class ships. The Pugliese design filled the volume of the TDS with a large cylinder, which was in turn filled with closed tubes reminiscent of those in HMS Ramillies. Pugiese's theory was that the torpedo would expend its energy crushing the cylinder. In practice the design failed miserably. Following the path of least resistance, the blast traveled around the cylinder and concentrated itself against the weakest point of the complex structure supporting the cylinder: the concave holding bulkhead.

This bulkhead acted much like a dam mistakenly built bowing downstream, rather than upstream against the current. This concave surface was structurally the weakest possible arrangement for containing the force of an explosion, and to make matters worse, the workmanship proved tragically defective. Conte di Cavour sank from a single torpedo hit at Taranto, and Caio Duilio had to be beached to prevent her sinking, also after one hit. Littorio suffered three hits, grounding her bow before she could sink. Vittorio Veneto twice, and Littorio once, suffered severe flooding in dangerous situations at sea when struck by torpedoes, more than such modern ships should have.

Pugliese's design also consumed tremendous volume, and foreshortened the depth of the armored belt, making the ships so fitted more vulnerable to shell hits below the waterline. Once again, practical experience proved that not every innovation represented an improvement.

The final innovation occurred in 1937, with the Japanese Yamato Class. In Yamato, the Japanese carried the internal armored belt all the way to the double-bottom to form an armored torpedo bulkhead. Although this armored bulkhead was substantially thinner than the belt armor it was joined to above, it was still very thick and rigid by comparison to the thin elastic bulkheads introduced by the Americans in 1915. Unfortunately, such bulkheads were too rigid and prone to displacement from their mountings, permitting flooding around them. Worse, in the Yamato's case, a poorly designed and constructed joint between the armored belt and the torpedo bulkhead proved prone to failure and drove its supporting structure backward, puncturing the inboard holding bulkhead. The US also employed this variety of TDS in fast battleships of the South Dakota and Iowa classes and came to the conclusion that the heavy bulkhead was too rigid, resulting in a modest down-grading of the system's explosive resistance rating.

There is possibly another hidden flaw in the modern layered TDS systems of the fast battleship era. Given that the TDS paralleled angled internal belts, this placed the upper edge of a bulkhead in the TDS nearer the explosion and the lower edge farther from it. This created the potential for greater stresses to act on the upper edge due to proximity, or on the lower edge due to channeling. Either effect would concentrate the force of the explosion against the weakest areas: the upper and lower attachment points of the bulkheads. By contrast, the 1915 vintage TDS of the Tennessee class placed all bulkheads parallel to the ship's vertical side, possibly ensuring the most even distribution of force across the face of the bulkhead.

Despite all of the design features intended to moderate the effects of a torpedo hit, the single most important factor in the effectiveness of a TDS remained its depth. The greater the distance between the point of impact on the side shell and the holding bulkhead, the more likely the system would protect the interior compartments. The French battleships of the Richelieu Class are often credited with the most effective TDS, but this is largely due to its extreme depth amidships. In other respects the design was very conventional.

Even in the Richelieu Class, the depth of the system was not constant from bow to stern, tending to taper and thin out toward the ends. This was also where it was least affordable: near the magazines. This factor also compromised the highly effective TDS fitted in the American fast battleships. The demands of high speed dictated that US fast battleships be very fine forward, thus restricting the depth available to the torpedo defense system abreast the forwardmost turret. This reduced the system's effectiveness, with the result that the flash from a torpedo's blast reached the forward magazine of USS North Carolina (BB-55). Flooding from the hit fortunately prevented a fire. The demands of high speed also dictated a complex stern structure that restricted and weakened the TDS in the South Dakota class aft.

Finally, there was absolutely nothing the TDS could do for the ends of the ship. Flooding the bow materially impacted the ship's mobility, forcing reductions in speed to prevent progressive flooding and / or tearing of the side shell. Hits aft endangered the steering gear and propellers, the Achilles Heal of every ship for which no satisfactory protective scheme has ever been devised. The best a designer could hope to do was protect enough of the ship's buoyancy to permit it to remain afloat with both ends flooded. This was a firm criterion of US designs, but one not followed by the Germans in the case of SMS Lutzow and the Japanese in the case of HIJMS Musashi.

The final form of torpedo defense was subdivision of the main underwater spaces of the ship. Many ships split their large spaces with one or more longitudinal bulkheads, prevented flooding clear across the ship from a single hit. However, this resulted in off-center flooding and serious listing. Other ships employed a cruiser style layout called "the unit system" which divided complete, self-sufficient "units" of machinery into compartments separated from each other by athwartships bulkheads. This system reduced listing moments, but permitted flooding across the ship (with attendant negative free-surface effects on stability) and allowed a single hit to knock out all of the machinery supporting one propeller shaft. Ships fitted with turbo-electric drive enjoyed a uniquely fine form of subdivision that produced numerous small machinery compartments at the price of greater structural weight.

Throughout the history of the dreadnought, torpedoes proved the number one killer of the type, more than justifying the effort expended by designers to limit their effects.