Intriguing discussion. So Jespersen assumes that in earlier times ‘for’ in ‘forty’ was short and different from ‘four’, just like ‘five’ and ‘fifty’. The difference in spelling presumably reflected that in pronunciation, which has now been neutralized in many dialects.
(But if so, what about ‘fourteen’? There are these regular pairs with shortened (weakened?) forms: thirteen/thirty, fifteen/fifty. Is it ‘fourteen’—but not ‘forty’—that’s irregular?)