I'm not sure I'd call it a bug.
What appears to be happening is that "A1" extracts the displayed (or possibly the maximum precision possible to be displayed) number in the cell, but this is compared to the (more precise) calculated and stored value in the cells including the one displaying 4.?...
Numbers displayed in cells on Numbers tables have a maximum (base 10) precision of 15 digits. Numbers that are the result of a calculation have, If I recall correctly, a maximum (base 10) precision of 19 digits.
Whether the displayed number is larger or smaller than the same number expressed to a precision finer by 10^4 depends on the direction it is rounded at the 15th digit. If the rounding is down, the effect is the same as truncating the number after the fifteenth digit—the added value of any digits after the fifteenth is lost, and the displayed number is less than the calculated value. This will always be the case for a positive value which includes 1/3 as a fractional part. For numbers including a fractional part of 2/3, though, the opposite is true: no matter where the repeating decimal representing 2/3 is rounded, the final digit of the rounded value will be a seven, and the displayed (rounded) version will always be larger than the calculation result.
A fractional value of n/7 will produce a similar result.
In this case, n = 1, 4, or 5 provided a value that was rounded down at the 15th digit, and was counted as being 'smaller than itself'; values of 2, 3 or 6 in the numerator rounded up, and counted as 'larger than itself.'
Regards,
Barry
