How can I tell whether a number shown in hex represents a byte or a character?

Those are different interpretations, and the display alone may not tell you which one is intended. 41 is a number written in hexadecimal. If the viewer labels it as a byte, it is one byte value; treating that byte as text requires an encoding. A text viewer might show a decoded character alongside the byte, but that displayed character is the result of applying an interpretation—not a second meaning contained in the hex itself. [S1]

For a practical first check, look at the column heading, display mode, or documentation for the tool. Does it say “bytes,” “hex,” “text,” or name an encoding? If all you have is 41, you can report the value as hex without guessing a character.

First, confirm that the values are bytes from a text source rather than numbers or other binary data. Then identify the encoding used to turn text into bytes. For some encodings, one character can involve multiple bytes, so you may also need the bytes before or after the value and the boundaries of the sequence. UTF-8 uses one to four bytes for a Unicode scalar value; UTF-16 can use a surrogate pair for a supplementary value, so a lone code unit may not be meaningful on its own. [S1]

A useful minimum is: the full byte sequence you are examining, the encoding if known, and whether the display starts or ends in the middle of a sequence. Byte order may also matter for the data being interpreted. If those details are missing, mark the character interpretation as unknown rather than presenting a guess as fact.

Yes. Keep the observation separate from the interpretation. For example, suppose a display shows 41 C3 A9. The following is a cautious note, not a claim that the values decode to particular characters:

Observed: 41 C3 A9 (hex values shown by [tool or source]). Interpretation: not confirmed as text. Encoding: unknown. Sequence boundaries: [known/unknown]. To identify text, confirm that these are bytes and obtain the encoding and complete sequence.

If you later confirm an encoding and a text decoder produces output, record the encoding and the exact bytes used. If the sequence could be incomplete, say so; a fragment may not be enough to interpret text reliably. Hex digits are useful evidence about values, but without that context they do not establish a character. [S1]

Name the level you actually have. A byte is the value shown in a byte-oriented dump. A code unit is a unit used by an encoding representation; it is not necessarily a complete character. A code point identifies a Unicode value, while a displayed character is what a text viewer presents after interpreting data. Don’t infer one level from another just because the same digits appear on screen. UTF-16 surrogate pairs are one example of why a single unit need not be a complete value. [S1]

When in doubt, write “hex byte value” for a byte dump and “decoded text” only when the encoding and byte sequence are established. The available information may still be insufficient to determine whether bytes were intended as text or to recover a unique text interpretation.