NavList:

N(x) is actually an logarithm of something. Assuming you are always using the same table the absolute value of the entries in a log table are irrelevant - only  their ratios matter. So if you where to divide your table by the maximum value - here 5840, actually 0.5440 - then the biggest number would become 1.0000 or, written w/o period, 10000. The other values would grow in the same ratio.

 If you continue using just 4 digits again (except for the entry 10000 itself) all of a sudden you get a resolution 1 out of 10000 out of the  4 digits.  Therefore, you practically doubled the available resolution and, accordingly, reduced any rounding errors by abt. 50%. This is significant particularly for the very small entries.
 

So, the idea now is doing just that: divide all entries by the biggest entry, and you will exhaust the entire number space available therefore minimizing possible rounding errors. Again, that number space when 4 digits are used is now 0 to 9999 in steps of 1.
D.. Burch could have used with advantage that scaling  for his original N(x) .

Only if you are concerned about communicating the logarithms per se and not the corresponding  natural value to somebody else there would be a problem, and you would also have to communicate the type of scaling you did. This is never a problem when you do SR with the same scaled log table.

BTW: the way such scaling is communicated is by mentioning the base of the logarithm like 2, e, or 10 which are actually being used in technology.  Here you don't even have to know that.