I looked at the TOSA test results (position and size of holes in the hull, etc.) and the Japanese diagrams of underwater trajectories and, interestingly enough, the hit seems to have been a PERFECT example of a tumbling pointed shell, in that it made a hole in the hull in the form of a slot from a SIDEWAYS impact. Thus, all of the design effort on the Japanese Type 88 and later Type 91 AP shells, with their break-away noses, lack of improvement in armor penetration ability over their immediate post-WWI British "Greenboy" design basis (British 13.5" and 15" Mark 5 APC shells of circa 1921-24), and super-long fuze delays, was based on a projectile with a DEFECTIVE FUZE, since the projectile did not detonate after moving a rather long distance through the water, but did detonate after punching through several plates, one of which was the heavier anti-torpedo bulkhead between the torpedo defense system and the hull spaces proper. If you look at the position of the explosion inside the ship, near the centerline, this corrolates with normal fuze delay action after punching through that anti-torpedo bulkhead set the fuze off. Thus, it is ironic that such a major effort was based on this cause, with so little so show for it during WWII.
Only once, to my knowledge, did the design work exactly as planned, with an 8" Type 91 AP shell hitting the main forward magazines of the cruiser USS BOISE near the bottom of the hull, well below the armor, which resulted in the burning out of the forward magazines, but not causing a major detonation since the Japanese did not count on the US pure nitrocellulose propellant taking so long to build up pressure in the magazine compartment hit, especially as the hole in the lower hull from the 8" Japanese shell was acting like a fire hose to douse the fire. With British and Japanese cordite propellants (nitrocellulose and nitroglycerine mixtures), the ship should have blown up like HOOD. Here was a failure to "know your enemy".
Kind of depressing to see how easy it is to have normally intelligent people go running off in a "wild goose chase" from so little evidence. I think some bigshot got a "bee in his bonnet" about that TOSA hit and eveyone else just went along because he was boss, not because he was right.
Also, shell weight does not matter very much as to penetration of face-hardened armor (it is only to the 0.2 power), though full weight effects occur against homogeneous armor. The homogeneous armor gives way slowly and the entire projectile momentum has time to get involved in punching a hole. Thus, projectile weight and the square of the velocity can be seen to balance the complete kinetic energy required to penetrate (a heavy projectile has a lower striking velocity to penetrate in accordance with the conservation of energy of the entire plate and projectile).
Face-hardened armor, however, is rigid and does not give until the force gets too great to resist, at which point the hard face caves in suddenly and the pressure can then tear through the soft, ductile rear of the armor. Usually, the face breaks prior to the full weight of the projectile getting involved -- the impact shockwave that reflects and cracks the plate face layer involves only the projectile nose and by the time the shockwave in the projectile reaches the lower body of the projectile, allowing that lower body to get involved in the penetration process, the face has already either broken, allowing penetration to proceed, or it hasn't and the rest of the projectile's weight is not going to help. Only in a narrow range of striking velocities does the full projectile weight get involved at all and this is where the projectile barely penetrates the ductile back layer after the face layer has cracked open, so hitting above this velocity renders the full projectile weight of no consequence. If the plate were hard all the way through, projectile weight would mean even less, once a minimum is reached.
This is why the use of 2700-lb 16" shells (US final Mark 8 AP design) does not buy you much compared to 2100-lb 16" shells (WWI Mark 3 AP design) -- a 29% weight increase -- against the same face-hardened armor plate, assuming equal damage to both projectiles from the impact. You only can reduce the impact velocity with the heavy shell here to 94% of what you need for the light shell; hardly worth the effort to make the heavy shell here.
Against homogeneous deck armor, there is a definite gain with the heavy shell, which (1) gets the full effect of the weight increase; (2) is fired at a lower muzzle velocity and thus has a steeper angle of fall at any given range, reducing the effect of oblique impact on the penetration and greatly improving penetration at long range (all else being equal); (3) has less air resistance effects due to more mass concentrated behind a narrow frontal area, increasing both the range and the striking velocity somewhat, so the loss in striking velocity compared to the lighter shell with a higher muzzle velocity is not as great at long range as one might think; and (4) has a larger explosive filler charge, since these usually are given as a percentage of the full shell weight, so a heavier (longer, usually) shell of a given basic design has a larger explosive cavity, too.
Thus, making a heavier, lower-muzzle-velocity shell has some advantages at longer ranges, though not much at close range.