I never ruminated about why the new pattern Teflon seals needed to be set at less depth than the spring-loaded seals, I just went ahead and did it because I am a slave to Authority, but after reading the typical BMWglish instructions umpteenth times, I've come to the following sleep-deprived conclusion. If the arbor for setting the old-style, spring-loaded seals were used on the Teflon seals, it could drive the seal's outside edge over the chamfer as indicated by the arrow. Whether that could happen was dependent upon the variation of depth in the block where chamfer were machined—a location that could be the "variation" in tolerances mentioned in the bulletin. The chamfer could deform a deeply-driven seal enough to allow oil migration between the seal and the block wall. Heat and oil pressure would eventually push oil completely past the seal, past the intermediate housing, and onto the clutch housing.

So BMW lessened the depth that the new-style seal was driven to account for variations in machining. The result is a new arbor was created to set the seal's outer surface at its current specified distance from the block surface.






IanT has a chance that his seal won't leak if it isn't low enough in the seal seat to deform on the chamfer. It's a calculated risk with reasonable odds and, like I mentioned previously, can make life interesting in anticipation of the outcome. If Ian takes those odds, a positive outcome would be useful information to provide the forum members.

Ian should also realize that he is not the first of us to screw up seal installation and that it is a comfort for many of us to know somebody else did it, too, so let's all thank Ian. 

 Thank you, Ian.

Was my memory good, or shitty?

Edit: lower instead of proud? Then it's more than just my memory that's shitty? I've been confused about that since when I first read it, that's one thing I do remember.

I'm running off a shakie memory here. I think some of the newer seals bottom out earlier. If it's the same as the original it's possibly ok.

I remember reading a quote about Bill Shakie on a dunny wall:

One would think,
With all this witt,
That Shakespeare himself,
Came here to shit.

Incidentally, you don't need a code reader, you can use a test light. Try following the methods described in this post.

The "test light" needs to be a single low power 12V LED. The diagnostic plug does not put out very much power so if you try something like a multi-LED 1156 turn signal bulb it probably won't illuminate.

Ha ha! No problem. It’s a K100RS though mate. Not a K1100 you’ve done that a couple of times and I just wanted to point it out so we’re not at crossed purposes.