The form structum seems to have "c" by analogy to stems that had labiovelar consonants in Proto-Indo-European, as Alex B. says.
In frūctus (from fruor) the Indo-European present stem ended in a labio-velar gʷ, but various analogies no doubt account for strūctus, flūxus, old flūctus, uīctum (from struo, fluo, uīuo) (Vox Latina, W. Sidney Allen, Second Edition, p. 69) (First published 1965, Second edition 1978)
There is also a relevant footnote on this page:
- Old Latin [for fīgo] is in fact fīuo, with u from Indo-European gʷ. But note nīxus (and nictare) from (co)nīveo, where u is from Indo-European gʷh.
In Latin and Greek Accidence, by John Perkins, I found the following examples of verbs that have a velar consonant only in non-present stems:
- fluo, fluxi, fluctum/fluxum
- struo, struxi, structum
- vivo, vixi, victum
(p 36)
- fruor, fructus/fruitus
- nitor, nisus/nixus
(p 37)
Wiktionary (accessed 24 June 2017) gives the following etymologies:
fluo: "From Proto-Italic *flowō" (cites De Vaan), "from Proto-Indo-European *bʰlewgʷ-, from *bʰlew-"
struo: "From Proto-Italic *strowō" (cites De Vaan) "with spurious c in struxī and structum"
vivo: "From Proto-Italic *gʷīwō [...] The x and c in vīxī and vīctum were introduced by analogy with other verbs." I left out what Wiktionary says about the Indo-European etymology because my understanding is that this is a somewhat troublesome root to analyze
fruor: "From the Proto-Indo-European *bʰruHg-" (De Vaan is listed in the "references" section, but there's no explicit citation and no Proto-Italic form given)
nitor: "From Proto-Indo-European *kneygʷʰ- (“to bend, to droop”). Cognate with nicō, connīveō, nictō"
As you might have noticed, Wikipedia, unlike Allen, doesn't write a macron on structum, and there is a note in the Wikipedia entry for struo saying:
Please note that there is a disagreement over whether or not there is a macron on the third and fourth principal parts and the subsequent verb forms from these (strūxī for struxī and strūctum for structum).
However, there is no citation for this statement, so I don't know what the basis for the disagreement would be.
The Latin language; a historical outline of its sounds, inflections, and syntax, by Charles E. Bennett (1907) says
Gellius [...] testifies (Noctes Atticae, ix. 6) to the quantity of the vowels of āctus, lēctus, ūnctus, and in xii. 3. 4 to that of strūctus (p. 52)
and
the Romance languages [...] point to cīnxī, distīnxī, exstīnxī, fīnxī, pīnxī, strūxī, tīnxī, ūnxī (p. 53)
but perhaps there are some issues with this evidence, or opposing pieces of evidence. Bennet doesn't indicate any inscriptional evidence (e.g. use of the apex) for vowel length in this word.
###Further reading
Some links to further literature that seems relevant that I found by Googling, but that I haven't processed yet:
"Latin Verbs in -uo, -uere", Oswald Szemerényi, in Italic and Romance Linguistic Studies in Honor of Ernst Pulgram (1980). This seems pretty detailed! I can only see part of it through Google Books, but it looks worth a read.
"On some Slavic words in -č: k from laryngeal?", by Jens Elmegård Rasmussen (1988)
"Lachmann's Law (Part 2)", by Roman Sukač (2013)
"On Latin strāgulum and strāges: -g- and analogy", by Kanehiro Nishimura, in Words and Dictionaries: A Festschrift for Professor Stanisław Stachowski on the Occassion of his 85th Birthday (2015)