>… it seems to me, that what is called `cavelength` in Therion is actually the length of all centerlines.

Yes

>This is for on the one hand a problem with circular shots inside big rooms, like http://filez.foxel.org/3s3H0d2X361A0K222Z3N

>An other problem are crossroads like http://filez.foxel.org/1t0C0r2Q011i0g1L0i00 where you can't young the shared area of two passages twice but must - at least to what I have been thought - >deduct some part of the centerlines: in the following picture I have marked the parts being considered cavelength yellow and the centerlines not counting as part of the cavelength >red https://img.skitch.com/20120221-jhraj625isy8j1i9naxiki3b76.png

I don’t get too concerned about this.  Like Torstein I figure that if it is similar to a length that you have to walk or crawl along to go from one passage to the next (crossroads), then it is part of the cave length and generally it is close enough to the centerline length as well.  Where there are legs that circumambulate big rooms or duplicate passage already surveyed, then I just add ‘flags duplicate’ (or maybe sometimes ‘flags splay’).  This removes these legs from Therions centerline cavelength calculation.  If one is concerned about precise lengths then one could ensure they place a survey station at each transition between what they deem to be ‘duplicate’ and ‘original’ cave passage.

Bruce