I was marveling today at the word hebdomadal, from the Greek ἑπτά for seven. But that had me wondering why words derived from seven sometimes use /bd/ and other times /pt/. I notice, for instance, that the word for seventh is ἕβδομος. (The same could be asked of ὀκτώ and ὄγδοος.) And among compound numerals, the cardinals persist in using /pt/ and /kt/, while the ordinals persist in using /bd/ and /gd/. What's going on here?

I consulted Smyth's Greek Grammar, and it distinguishes in §16 between the order of stops—/p/ and /t/, as unvoiced stops, belong to the first order, while /b/ and /d/, as voiced stops, belong to the second—and per §82, a labial or palatal before a dental changes to the dental's order. Smyth lists ἑπτά as an example in §82.c. N. 1, so I think I'm on the right track. It appears the dental is changing from smooth to middle order, and the labial with it—my question is, why?

  • 1
    Good question. I really don't know under which circumstances a stop would change from voiced (dgb) to unvoiced (tkp) or the converse. They do that when followed by another stop, as in the the p and the b in this case; but why would the t and the d be different, as they must come from the same root? When did they differentiate, and by what rule can this be explained?
    – Cerberus
    Oct 14, 2019 at 22:34

1 Answer 1


The Greek word for seven, hepta, like Sanskrit sapta, Latin septem, and others, points to an Indo-European *septm. In Greek *s becomes h, and the syllabic *m becomes a. In the word for “seventh”, IE *septm-o-, the m is a consonant (non-syllabic) and remains m in Greek, giving *heptmos, and then (with voicing of pt to bd before the voiced m) *hebdmos, and then (with insertion of a vowel to break up the cluster) hebdomos.

  • Thanks so much! I have to ask, what's happening with ὀκτώ and ὄγδοος, then? Is it something different? I see the PIE was *oḱtṓw or *h₁oḱtṓw—the long o didn't drop out, and no stop follows it, anyway. Analogy, maybe?
    – Nick
    Oct 15, 2019 at 13:40
  • 1
    @Nick. “Analogy” is the easy way out, but as yet no one has suggested anything better.
    – fdb
    Oct 15, 2019 at 19:21
  • 3
    The explanation I've heard for ὄγδοος is *oktwos > *ogdwos > ogdowos, exactly parallel to ἕβδομος, with voicing assimilation to w. I don't know enough about the morphology of PIE ordinals to say whether the absence of ō is expected.
    – TKR
    Oct 15, 2019 at 23:57
  • 1
    @TKR. Yes, this and other proposals are discussed by Beekes, s.v. ὄγδοος. In this case it is the assumed suffix *wo that is problematic.
    – fdb
    Oct 16, 2019 at 9:59
  • More precisely, that proposal doesn't necessarily assume a suffix *wo, but a -w as part of the stem. But it all seems very unclear.
    – TKR
    Oct 20, 2019 at 0:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.