Just to be devil’s advocate, he did massage his results somewhat to fit his ratios. Having said that, his ratios work out for traits that are truly monogenic, it's a pity these traits are not the majority. Chances are some of the original traits he tracked were not simply present/absent but exist in a continuum; he no doubt made a judgment call about the categorisation of some individual. But he used large numbers and consciously or not compensated for the deviation from his ratios by the slightly "less" monogenic traits. This is probably the mark of a good scientist, has a sound theory based on a sound logic (ie genes = units of trait heredity) rooted in the physical world, and the intuition to realise that the deviations that arouse weren't lethal to the theory. And he was right, albeit on a simplified level. Nowadays there is an acknowledgement that most traits are quantitative (ie non-monogenic).
OK how about applying mendalien genetics to sex determination?
In 1927 K. Hirata at the Hokkaido Imperial University published work where he induced hermaphrodite formation in Cannabis sativa by mutilation, isolated each plant so it only fertilised itself, allowed the buds to mature, collected the seeds and replanted them. 294 seeds were collected, 143 germinated of which 78 were female, 45 were intersex (AKA hermaphrodite) and 3 were male, the rest never reached sexual maturity. (Hirata, K. Journal of genetics (1927), 19(1): 65)
Hirata postulated a sex chromosome setup analogous to our XY system, and Mackey (and maybe other since) visualised homologous pairs of chromosomes for all but one pair, suggesting the existence of the semi-homologous XY pair.
One of the observation Hirata made was there seemed to be a range of intersex phenotypes which would argue sex as a quantitative trait.
All of this sprung from a discussion I had with someone about the generation of feminised seeds, got me thinking about exactly how you would produce a feminised seed. My simple brain thinks that the seeds from a selfed hermaphrodite would always be female, but this is not the case, and this argues for the contribution of multiple genes in the determination of sex.