Out of 46 mode 3 responsories (restituted so far) that are on (c4) (some are on other clefs and could not be included in the stat but those are very rare), we have 29 times si-sol, 7 times la-sol, and 10 times another formula (proper verse, or second part too short making it begin with its end formula).
The 7 la-sol are as follows: 1024N1R3 (inauthentic), P0F3R1 (from Liber Hymarius), 1007N1R2 (inauthentic), APTPN2R2 (from LH), P7F2R2 (verse only is synthetic, but the original verse does not have an incipit formula for its second part), P7R1 (from LH), and P6N3R2 (restituted verse from the franciscan sources). Clearly, GFS has si-sol consistently, and la-sol merely reflects the result of the work of Sandhofe and the authors of LH (and before them, Gajard et al).