More on what Alan mentioned.
Note that no ladder instructions are added to the program for this example.
I have an ML1100 program that uses 865 instruction words and 702 data words. It contains user data files 3 through 13.
Now I add file 14, configure it for integer, and give it a single element. After program verification, I see that I have used 703 data words as expected, but the addition of the file also used two instruction words, so now I have 867 instruction words.
If instead of designating the new single element integer file as file 14, I decide to skip one file and make it file 15. I will use three instruction words (868) and one data word (703).
If I decide I want the new file to be N100, then after program verification I will have used 974 instruction words while the data table words used is still 703. Skipping to N100 costs an extra 107 words.
And if I make the new file N255 instead, then I will use 1168 instruction words, and still only 703 data table words.
Most of the time you probably won't care and grouping the files by function can make a program more readable, for example, zone 1 might use files B13, T14, C15, N17, and zone 2 will use B23, T24, C25, N27, etc. But if you ever do run into a situation where you are running low on memory then making the files sequential can recover some much needed instruction words.