Passthrough mode - X-Bogosity header & its effect on the corpus

[Stroller]

Hi,

I've recently installed Bogofilter & given it some basic tuition using 
my spam & personal (ham) folders. I reckon this is probably a slightly 
dumb question, as others must have considered it before, but I can't 
see it covered in the FAQ or README anywhere.

I use maildrop as my MTA & I wish to use it in conjunction with 
Bogofilter to move all bogus-looking messages into a "Spam/Probable" 
maildir. I intend to review this folder periodically using my mail 
client, removing any ham & moving spam into a separate "Spam/Definite" 
folder, so that I can use a monthly cron job to teach Bogofilter 
further.

My concern is with the standard maildrop recipe that is supplied with 
Bogofilter, and with Bogofilter's passthrough option.

   xfilter 'bogofilter -f -p -u -l -e -v'
   # if spam, file this into the "spam-bogofilter" folder and exit 
processing
   if (/^X-Bogosity: (Spam|Yes)/)
   to "spam.probable"

Every spam in my probably folder will then have an additional header 
line something like this:
   X-Bogosity: Yes, tests=bogofilter, spamicity=0.999127, 
version=0.13.7.2

My concern is that when I move this message to my "Spam/Definite" 
folder & use it to teach Bogofilter, then that header will affect the 
corpus. I guess there are a number of possible repercussions - that 
Bogofilter will expect an incoming message without the "X-Bogosity: 
Yes" line to be ham, that the word "Yes" has an increased bogosity, or 
just that my Bogofilter database will get filled up with 
"spamicity=0.x" entries.

Does Bogofilter ignore X-Bogosity headers..? I can see no mention of 
this in the manpage.

Thanks in advance for any advice,

Stroller.