We discuss the evolution of binaries with moderately high masses (about 10 -
30 solar masses), and with periods of about 3 - 300d, corresponding mostly to
early Case B. These are usually thought to evolve either by reasonably
conservative Roche-lobe overflow, if the initial mass ratio is fairly mild, or
else by highly non-conservative common-envelope evolution, with spiral-in to
short periods (hours, typically), if the initial mass ratio is rather extreme.
We discuss here a handful of binaries from part of this period range (about 50
- 250d), which appear to have followed a different path: we argue that they
must have lost a large proportion of initial mass (about 70 - 80%), but without
shortening their periods at all. We suggest that their behaviour may be due to
the fact that stars of such masses, when evolved also to rather large radii,
are not far from the Humphreys-Davidson limit where single stars lose their
envelopes spontaneously in P Cygni winds, and so have envelopes which are only
lightly bound to the core. These envelopes therefore may be relatively easily
dissipated by the perturbing effect of a companion. In addition, some or all of
the stars considered here may have been close to the Cepheid instability strip
when they filled their Roche lobes. One or other, or both, of high luminosity
and Cepheid instability, in combination with an appropriately close binary
companion, may be implicated.