John,
The easiest way to answer your question, without having you read everything posted here, would be this:
The numeric figure of a poss het comes from the potential outcome you get from looking at the results of a Punnett square. Specifically, it is the statistical number of animals that could be gene carriers divided by the total number of animals that are phenotypically WT.
So if I give you an animal and tell you that it is poss het (I am intentionally not giving you a value) how would you set up the Punnett square for breeding that animal to a WT?
Because gene pairs are binary, you set up the breeding as a 2x2. So your Punnett will look like this:
And again, because genes are binary, ultimately there are only two possible outcomes here, either 50% WT and 50% het or 100% WT and 0% het.
The reason you cannot have a 25% poss het is because, due to the binary nature of the genes, there is no way you can take a 2x2 Punnett and get an outcome where only one of the four grids is het while the other three are WT. Likewise, you cannot have a 33% poss het because you absolutely cannot get 1 1/3 of the four grids to be het.
.
.
Or, if that is still confusing, perhaps I can try it this was. The snake itself is not 66% het or 50% het or any damn% het. The snake either is het or it is not het. The percentage is based on the statistical chance you have of picking an animal that is het. So, when you finally get around to breeding that snake, you have to set up a Punnett square to determine statistical odds. But again, at the end of the day, the snake either is het or it is not het so you must ignore those values when setting up the square for this breeding because the animal either has the gene and passes it on or it does not have the gene and does not pass it on.
Trying to put the values into play is like doing some Schrödinger’s cat game where the animal exists in both genetic states at the same time where the animal has the gene but actually does not but only 2/3 of the time (or only 1/2 of the time.)