It's understandably assumed that the word for 'ten', *deḱm̥, and the root used for the higher decads, *(d)ḱomt-, are etymologically related and so this is fundamentally why *d is added, however this voiced stop doesn't actually show up in the linguistic data! The only thing that shows up in its place is an added length in the preceding vowel, if anything. For example, since in Latin we have quadrāgintā '40', it's presumed that the first long vowel that precedes the decadic ending -gintā is from a lost *d. Hence, some presume the reconstruction *kʷetwr̥-dḱomt- with the implied intermediate form *kʷetwr̥̄-ḱomt-. Proponents of the Glottalic Theory, a variant of the standard account of PIE, assume even further that if *d were really an ejective *t’, then it would make sense that it would have simply eroded to a glottal stop. Once the glottal stop disappeared it would have lengthened the previous vowel due to compensatory lengthening. Given my Hybrid Theory, I could probably get away with the same explanation but so far, I'm having trouble buying these ideas even though they seem to explain the long vowels nicely enough. One would think that if *d really were present in these words in PIE, at least one language would retain it, no? Fishy. Very fishy.
To me, it begs the question: Are these IndoEuropean specialists strictly reconstructing PIE here or are they reconstructing some idealized pre-IE stage and then repackaging it as though it were PIE itself? I would like to suggest something new, if anything, just to be cheeky.
What if the underlying forms for the higher decads are instead inanimate compound words with the first element of the compound marked in either the attested dual ending *-h₁ or collective plural ending *-h₂ to explain the "compensatory lengthening"? And what if the second element is simply the root that we see attested, *ḱomt-. Thus:
*wih₁-ḱm̥tíh₁ ~ *dwih₁-ḱm̥tíh₁ '20'Perhaps these compounds then could be fossils of entire phrases in an older stage. Perhaps something like the following in mid Late IE:
*(d̰waiʔ) km̥tíʔ 'two tens'I think this accounts for the data better and may help us move forward with the origins of the IE numerals because it's simply not necessarily the case that the decadic ending and the word for 'ten' must have exactly the same consonantism. That's an untested assumption that isn't directly observable from the evidence. For example, we have no way of knowing just by looking at the IE data if *de- in *dékm̥ was once a seperate morpheme or whether the two roots are even related at all.
*treiχ kámtχ 'three tens'
*kʷetwárχ kámtχ 'four tens'