Some similarities are known to be utterly irrelevant to the hypothetical analogy (e.g., swans grouped by color is irrelevant to whether congruent geometric lines are grouped).

Yet it would be absurd to find positive support from this analogy for the idea that we are likely to find congruent lines clustered in groups of two or more, just because swans of the same color are commonly found in groups. The positive analogy is antecedently known to be irrelevant to the hypothetical analogy. In such a case, the analogical inference should be utterly rejected. Yet rule (5) would wrongly assign non-zero degree of support. To generalize the difficulty: not every similarity inc