Files contain 4 random hex characters

I see erroneous strings of four hexadecimal characters in generated .html files.
In about half of occurrences, these strings appear as the first four bytes of the file. In the other half, they are randomly placed in word breaks throughout the file. (That is, I see stuff like </td>301e<td> but never stuff like </td><t301ed>.)

/my/newly/generated/static/content/ $ find . -iname \*.html \
  -exec grep -P \
  '\b(?!2600|2500|2277|2000|2660|200[0-9]|201[0-7])[23]{1}[0-9a-fA-F]{3}\b' \
   {} -o \;                                 
2360
2563
20a9
30d4
2ac0
20ad
2ae7
3e6b
245d
2286
209d
2dbc
2024
2ac0
2843
390c
201c
2e9c
201d
2ae5
222c
3935
22e4
2f89
242a
20a6
2286
2151
...<826 lines total>

All the strings start with ~2″ or “3”, with the following three characters being hex.