it's not - because bolts move!  that's why there's a whole industry
dedicated to the production of locking mechanisms for threaded fastners.
 usually, we only care about the ones that loosen because they are more
likely to cause the failures.

which is consistent with lash of the pulley wheel!

that's part of it, but ultimately, as we can see from the skid marks on
the underside of the washer, the bolt is still free to turn.  the
question is, at what torque.  once it is turning, that sob is still in
there /way/ tight, and way tighter than when torquing to fastening spec.

depends if we've spent time doing this kind of work before!

I still dunno. I've seen galled steel flat washers and bolt heads various
places before (but I don't recall where they have been), so I wouldn't have
thought it was unusual. I would speculate that the torque we need to apply
to break crank bolts loose isn't being directed to the threads but to the
head, where the galled surfaces are responsible for the excessive
break-loose torque. Pure speculation, though.

Usually crank bolts (Honda or otherwise) need to loosen a quarter turn or so
before they come free, and then there is no evidence of the threads
galling - leading me to the speculation of the galled head and washer
surfaces being the key. That would also be consistent with the observation
that the break-loose torque goes up over the years, if engine heat and/or
vibration is important in the development of the galling.

I don't think we have enough to work with to come up with a definitive
answer.

Mike

ok, let's try this instead:

http://www.flickr.com/photos/38636024@N00/

lower res pics are here:

http://www.snapfish.com/thumbnailshare/AlbumID=31395672/t_=36454773

this afternoon, i went to my favorite junkyard and bought two crank
bolts.  one from an 91 civic, one from a 92 civic.  i'm going to post
the pics later this evening, but the observations are these:

_91_
* eyeball=poppingly hard to shift - had to get a fulcrum and bounce full
bodyweight at the end of a 18"x3/4" breaker bar.
* no evidence of loctite.
* clear fretting damage on the mating surface between the washer & the
bolt head.
* no evidence of corrosion.  [i'm in california]
* pulley wheel locked with single woodruff key.

_92_
* it was definitely snug, but i could remove with one hand.
* bolt thread clearly loctited.
* no evidence of fretting.
* no evidence of corrosion.
* pulley wheel splined /and/ woodruffed.

now, we all know what loctite does - it binds threads so they don't
move.  no movement means no possible further tightening.  loctite also
means a bolt is hard to remove compared to its fastening torque.

conclusions:

1. there is /definitely/ lash in the 91 pulley wheel - something that
honda evidently felt needed to be addressed with the addition of a
splined interface for the 92.  [splines don't eliminate lash, but help
mitigate it.]  fretting [or lack thereof in the case of the 92] is as
clear an evidence of lash as you can get.

2. loctite /prevents/ further tightening of the bolt!  hence the 92 was
much easier to remove, despite the loctite's binding function.  the
reduced lash would help in this regard also.

time to get out the camera...