The full count
Each bar is one of the 32 legal symbols, counted across all twelve addresses. A flat draw over 672 characters would put every symbol at 21.
Reading the spread
The counts run from 12 to 35 against a flat expectation of 21. That looks like a wide spread and it is not. With 672 draws over 32 symbols, the typical wobble around 21 is around four or five, so a range from the low teens to the mid thirties is ordinary. It is the same lesson as the digit page in a different form.
What would be worth noticing is a symbol missing entirely. That happens constantly in individual addresses, where six or eight of the 32 are typically absent from any one string. Across the full set it does not happen at all. Every symbol appears, which is the expected outcome and confirms nothing was filtered out of the alphabet.
The other thing the table quietly confirms is that no illegal characters got in. There are no rows for 0, 1, 8 or 9, and no rows for uppercase letters, because none appear anywhere in the 672 characters. That is the alphabet check stated as a table rather than as a pass mark.
Why frequency is not a fingerprint
People sometimes assume a frequency profile identifies a set the way a signature does. It does not, for the simple reason that it is a summary. Reordering all 672 characters at random would leave every count identical while producing twelve completely different addresses.
Frequency throws away order, and order is where all the information lives. This is why the record pages report frequency as a field but never treat it as identifying anything, and why the checks that matter on this site compare positions rather than totals.
Questions
Should every symbol appear the same number of times?
Only on average and only over a large enough sample. Over 672 characters a range of roughly ten either side of the expected 21 is normal.
Does a missing symbol indicate a problem?
Not in a single address, where several are usually missing. Across the whole set every symbol appears, which is what a set this size should look like.