Is it just me or is that last picture of the alpine butterfly wrong? The two turns in the middle should be crossing each other.
I'm unsure of what you mean by "two turns in the middle", but the middlemost
parts, which are SParts before they make an initial turn/curve, should always lie
parallel. The
collars of the resp. parts
seem to be a single part and so run
together, but that's an illusion.
Now, there is a tendency with the other commonly presented tying method--that of
making two half-twists and then bringing the bight tip around and through the center of
the first twisted rope--of generating torsion which will tend to make the legs of the eye
cross within the knot. This crossing appears to give the SParts more gentle curves at
their initial bend, and so might produce a stronger knot. But as the knot is asymmetric,
the testing of it needs to consider not only the initial orientation but also which end
is loaded as the SPart for making a loopknot. There might also be some change
in performance depending on whether the knot is first loaded one way, then tested another
way--e.g., load it qua loopknot in one direction a few times, then test its strength
in bend-loading.
Btw, the images on this site are nicely clear and well-shot. I'm glad that solid-color
ropes (vs. the patterned variety of kernmantle) were used, esp. w/bends.
*knudeNoggin*