(I believe a taconite pellet is a small black ball abt the size of a malted-milk ball. They would pack well.)

Even if "perfectly packed" in hexagonal close packed or body centered cubic configuration, identical spheres cannot occupy more than 74% of the total volume. (Ref for example See the paragraph right after equation 5.)

While I'd by no means assert that water taken on was a significant effect in this sinking, at least 30% of the hold volume would be available to infiltration unless someone placed the "balls" perfectly (one at a time by hand?). Since water would be significantly less dense than the ore balls, an increase of gross load by perhaps ~10% (+/-?) would not be impossible by taking on water.

Any likely shape other than perfectly spherical (e.g. like a charcoal bisquit?) probably would leave even more available free space in the hold, even if it "looked like" it was very densely filled.

John