A specimen is labelled 0042. Another is labelled 0108. The characters look numerical, but adding the labels would not tell you the combined size, age or weight of the specimens.
Ask what the column does
A measurement records an amount on a defined scale. An identifier helps distinguish a record or object. Before calculating, read the column description and ask which job it performs.
In this invented example, the leading zeros are part of the recorded code. Keep the identifier as supplied unless the dataset’s documentation gives a justified conversion rule.
Look for accidental interpretations
Sorting identifiers may help organise a list, but the order alone does not establish a biological progression. A larger-looking code does not necessarily mean an older specimen or a later collection date.
Similarly, counting records is different from adding their codes. If three records have identifiers, there are three records regardless of the numeric values their labels resemble.
Write the role into the documentation
A useful note might say “specimen identifier; not a measured quantity; retain formatting”. Include how the code connects to other tables if that information is available.
Before combining files, revisit the dataset checks. Understanding the role of a column prevents arithmetic that is technically possible but meaningless for the question you want to answer.