i believe the first two bytes are the number token bytes in the program, and then the rest is the data of the program.

This File Format Guide documents it, along with all the other TI file formats.

On a calculator, it's what nemo said- the first two bytes are the length, then it's followed by the data.

This is the text of the guide (minus HTML formatting):

The TI-83+ variable file format
The TI-83+ variable file format has two parts: a header and several variable entries.

The header appears at the beginning of the file and takes the following format:
Note - All 2-byte integers are stored little-endian Intel-style (least significant byte first).

Offset Length Description
0 8 bytes 8-character signature. The signature is always "**TI83F*".
8 3 bytes 3-byte further signature. These three bytes always contain {1Ah, 0Ah, 00h} = {26, 10, 0}
11 (Bh) 42 (2Ah) bytes Comment. The comment is either zero-terminated or padded on the right with space characters.
53 (35h) 2 bytes Length, in bytes, of the data section of the file. This number should be 57 (39h) bytes less than the file size.
55 (37h) n bytes Data section - consists of a number of variable entries (described below).
55 (37h)+n 2 bytes File checksum. This is the lower 16 bits of the sum of all bytes in the data section.



Each variable entry follows this format:

Offset Length Description
0 2 bytes Always has a value of 11 or 13 (Bh or Dh).
2 2 bytes Length, in bytes, of the variable data.
4 1 byte variable type ID byte (see variable type ID's)
5 8 bytes Variable name, padded with NULL characters (0h) on the right.
13 (Dh) 1 byte Version. Usually set to 0 (present if first bytes are Dh).
 
14 (Eh) 1 byte Flag. Set to 80h if variable is archived, 00h else (present if first bytes are Dh).
15 (Fh) 2 bytes Length, in bytes, of the variable data. (This is a copy of the value in offset 2)
17 (11h) n bytes Variable data. Click here for variable data formats.

Quote from: nemo on November 13, 2010, 04:33:02 pm

i believe the first two bytes are the number token bytes in the program, and then the rest is the data of the program.

When I look at the binary/hex code for an .8xp file on the computer, there's more at the beginning than that.

On the calc it just the two size bytes and a mess of data, though.

Quote from: Deep Thought on November 13, 2010, 11:48:53 pm

Quote from: Hot_Dog on November 13, 2010, 05:00:16 pm

Quote from: nemo on November 13, 2010, 04:33:02 pm

i believe the first two bytes are the number token bytes in the program, and then the rest is the data of the program.

When I look at the binary/hex code for an .8xp file on the computer, there's more at the beginning than that.

On the calc it just the two size bytes and a mess of data, though.

* Deep Thought wonders what this is for

probably for Correlation project. 

Exactly.  I just finished turning bmp fonts into .8xp files

If you don't mind using C#, I have a library for modifying all sorts of files, including .8xp.  Let me know if you are interested.

Sure thing.