The Hermie Gene (H) is dominant. However, activation of the gene is possibly dependent on another genes activation (L). This would explain the different activation methods of the H- pheno, environmental or otherwise.
If your plant hermies without stress it would be the H-L- phenotype. (My Durban Poison is this pheno)
If it hermies under stress, it would be the H-l- phenotype. (My Cordiero is this pheno)
If it doesn't hermie under stress, it could be considered a "pure female" and would be hL or hl. (My NYCD is this pheno)
When Mendel allowed his peas to self pollinate, he got the 3:1 ratio consistent with an F1 hybrid. His peas are monecious, cannabis is dioecious so the comparison is not necessarily valid, but you could assume you'd get the same results with a hermied cannabis plant. If the parent was also a self pollinated hermie, the chances of a hermie are higher in the progeny than if the plant is an F1 cross of a hermie and a non hermie...
If a hermie pollinates another one of your plants, the results will be a Dihybrid F1 Cross of the two gene pairs. The initial ratio of 3:1 would apply to the H- pheno and the L- pheno, together giving you a 9:3:3:1 ratio, with the highest probability being that the progeny expresses the H-L- pheno, seconded by the H-l- and h-L- phenos. Neither of which are what you wanted. This type of set up will "drift" towards the dominant gene pair (H-L- or "always hermie" if the line is followed.
There is, still, the minor chance (1:15) that the recessive h- pheno will be present in your seedstock and you will have successfully bred Hermaphrodism out of an F1 hybrid of the two strains, which can be backcrossed with the original "pure female"to begin a line breed of the two strains, free of hermaphrodism, which can subsequently be used to backcross with the original hermaprodite to produce new combinations of genes within the lines.