It is an area of light or see-through emulsion in each case, so that rules out static-sparks (and there wouldn't be so much anyway, and probably not at all in 135). Is there any visible surface imperfection on the emulsion? The different densities and sizes are very unusual!
As you are using chemicals made up from powder, can we just confirm that every batch is filtered into the measured container before use? (eg. fill the measuring cylinder(s) through a funnel with some washed cotton-wool in the spout).
Are the films showing this effect left in the tank 'dry' for any length of time? (perhaps allowing dried fixer-residue to fall in to the loaded reel? or using poorly washed and not-quite-dry metal reels?)
And the obvious one, if these are from a scanner, or contact-frame print, is the glass clean?! (though that would usually be seen as light specks or marks, not dark ones).
If there is no physical damage, it looks as though something is slowing development on those small "Y" areas - but what . . . and that is your question of course! Oops.