TABLE 3

Results of maximum-likelihood analysis of the spermicidal-chromosome model

With Df(l)bb452Without Df(1)bb452
H2 vs. H1G = 668, 16 d.f., P < 10–10G = 518, 14 d.f., P < 10–10
H2X vs. HIG = 329, 8 d.f., P < 10–10G = 219, 7 d.f., P < 10–10
H2Y vs. HIG = 282, 8 d.f., P < 10–10G = 266, 7 d.f., P < 10–10

Maximum-likelihood analyses were done for the hypotheses outlined in Figure 5. Initial analysis (see text) indicated that the data for bb452 are inconsistent with this model, but its disjunction is so nearly normal that it may be an outlier. The analyses were therefore done in two fashions. In one case, survival of X-bearing sperm was allowed to exceed 1 (an exception that affects only bb452). In the other, the bb452 data were excluded. Whether one evaluates a hypothesis in which both unpaired sex chromosomes are spermicidal (H2) or the two hypotheses that asked whether this might be true for only one or the other chromosome (H2X and H2Y), there is a large difference between the predictions of the spermicidal-chromosome model and the observations, and the probabilities of getting differences this big by chance are exceedingly low.