74 characters, but 203 bytes when they're UTF-8 encoded. My guess is that someone wasn't careful with (or had some technical reasons for) counting UTF-8 characters versus counting bytes, and taking away the last character brings it to 199/200 bytes and it fits into the field again.