Three prior "accidental" drownings are enough to show or suggest that the latest drowning was no accident.
Two prior drownings are also enough to show or suggest that the latest drowning was no accident.
So one prior drowning is also sometimes enough? (If not, why not?)
And, of course, no prior bathtub drownings are necessary and evidence of the putatively accidental but possibly murderous drowning for which accused is on trial is admissible to show that the drowning was no accident.
In short: if it's all about the numbers, there is a sense in which the numbers don't really matter. In any case, what does the number of prior drownings -- one, two, or more -- have to do with the question of whether an inference from propensity is necessary? The answer, I think, is nothing -- unless, that is, you think -- incorrectly -- that the character evidence rule is rooted in nothing except judgments (or beliefs) about the probative value of evidence about human behavior.
coming soon: the law of evidence on Spindle Law