I was doing the exact same math as you a few minutes ago.
https://en.wikipedia.org/wiki/Random_close_pack
This is the article I used to find out a packing density that would be reasonable for spheres in a container. (Judging by your 74% figure, you probably used the same one. Oh well, here it is for everyone else.)
I was assuming that they were being poured in from either a hopper or a loading station and used the low end of efficiency for poured random packing, which is 60.9%. The volume of one ball is 65.54 in^3, that divided by 0.609 is effectively ~107 in^3 per ball. The rest of the math should be fairly straightforward, and I got ~320 balls in your 40x36x24 container.