>On Lachmann's Law
>That there is no
>*pervasive* lengthening before, say, suffixal /t/ in roots ending in
>mediae is clear from the counterexamples which are very solid: fissus,
>sessus, scissus. Interestingly, tussis 'cough' has short u, while
>the ptc. tu:sus 'hurt, struck' has /u:/.
(1) tu:sus probably gets its length along with the alternative form,
tu:nsus, with regular lengthening before /ns/. The usual pattern in Latin
is -Vns > -V:ns > -V:s and sometimes reinsertion of the -n-. Therefore both
tu:sus and tu:nsus are expected outcomes of short u before ns, even if this
ns < *nd + tos. Compare the forms from tendo: here the alternatives are
tentum (short vowel!) and te:nsum (vowel lengthened before ns).
(2) Your counterexamples are all in -ss-. These cannot be persuasive,
because of the variation in early inscriptions between -Vss- and -V:s-.
Whatver the origin of these forms, analogical pressure was at work during
the recorded history of Latin pushing some forms to -Vss- and others
to -V:s-.
(3) Other counterexamples can be found instead, for example pingo pictum,
stringo strictum. These are less numerous than the forms in -s-/-ss-, so
its a shame the -s-/-ss- forms are not totally reliable.
Peter