On the inaccuracies which probably exist in the census returns of ages