Abstract
Analyses of secondary structures proposed for ribosomal RNA's show that, of the different kinds of base pairs directly adjoining the ends of postulated double-helical regions, only A-G with A at the 5' end significantly exceeds the number expected for a random base distribution. An A(syn)-G(trans) hydrogen-bonded basepair is proposed. This could fit at the end of an undistorted double helix, but would prevent further base stacking, thus favoring a break in the double helix to produce a non-linear tertiary structure.