The 72 is optimised, it seems to me, for annual compounding with percentages from 5% to 10% (as ln(2)/ln(1.08)*8 = 72.051746736), while, as you point out correc

The 72 is optimised, it seems to me, for annual compounding with percentages from 5% to 10% (as ln(2)/ln(1.08)*8 = 72.051746736), while, as you point out correctly, ln(2) = 0.69314718, so for continuous compounding 69 would be better indeed.