Missing 10nm in measures of shadowed SP-D hexamers vs multimer —
If using just one (so far) multimer, shadowed with tungsten (in the technique of J. Griffith’s lab), shows that the N terminus of SP-D to the CRD is about N=7 Mean=63.9, SD=4.02 for a single arm (trimer), and closer to N=3, Mean=138.69nm, SD=5.19nm which leaves a difference of about 10nm between 2x the trimer and the full width (diameter) of SP-D. This difference and the look of the bunch of N termini in the center of the multimer lead credence to the side to side (at least occasionally if not always) of the N termini between (among) hexamers of SP_D.
Just a good look at the center of this multimer causes one to believe that these N termini are arranged in a ring, not end to end, and just mechanically, how would end to end work anyway.

Light yellow line show the course of the plot (made in ImageJ) in each of the images. (Images are shown in inverted contrast). Trimer plots are N terminus on the left side, hexamer plots are shown with the N termini in the center (both are marked with a red N and also in the multimer, a red arrow).  Counting the peaks in shadowed images is problematic,  but in fact comes up just proud of what shows up for AFM images,  background peaks can be used to reduce the noise, but image processing has not been used here.