Experimental evidence for very long term processes

True enough. Remine however was extrapolating from a rate of substitution of 1 per 300 generations, which Haldane posited as the average tolerable maximum. Here’s the original paper: