Complicating our investigation into why Matthew has 41 generations is that some manuscripts have 42 generations in the genealogy. In particular, at Matt 1:11 some manuscripts insert a generation for Joakim between Josiah and Jechoniah. This reading, however, is rejected on both external and internal grounds, as Metzger explained in his textual commentary (p. 2):
In order to bring the text of Matthew into harmony with the genealogy in 1 Chr 3.15-16, several of the later uncial manuscripts (M U Θ Σ), as well as a variety of other witnesses (including f1 33 209 258 478 661 954 1354 1604 syrh with *, pal geo), have added τὸν Ἰωακίμ, Ἰωακὶμ δὲ ἐγέννησεν. Although it is possible to argue that the clause had accidentally fallen out during transcription, the external evidence in its favor is not weighty as that which supports the shorter text (א B C E K L S V W Γ Δ Π most minuscules it vg syrc, s, p copsa, bo arm eth). It should be noted also that when the clause is present there are fifteen generations in the second tesseradecade (compare ver. 17).
I agree with the decision to exclude the longer reading here due to the external evidence, but I have some issues with how the internal evidence is evaluated, especially in Metzger’s last sentence. It seems more likely to me that scribes may be more concerned with harmonizing the genealogy with the counts in verse 17, not against the one in Chronicles, because otherwise we should more witnesses restoring the three generations of kings. Rather, the goal is to add a generation to the second tesseradecade and transfer the final member of that second set to the third tesseradecade, so that there fourteen generations without double-counting in each of the three tesseradecades.
This approach may be older than the external evidence quoted by Metzger. It is known from Amy Anderson’s study of manuscript 1582 that the text family 1 has its antecedents in Origen. If this reading was present in Origen’s text, it may explain his assumption in his Homily on Numbers 27,3 that there were in fact 42 generations without any awareness of the counting issue.
Even older than Origen is Irenaeus, who is even cited in the NA 27 apparatus in support as Irlat vid. This is a reference to Irenaeus, Adv. haer. 3.21.9 which states Ioseph enim Ioachim et Iechonias filius ostenditur, quemadmodum et Matthaeus generationem eius exponit (“For Joseph appears to be the son of Joachim and Jechoniah, as Matthew too expounds his generation.”). The siglum vid (from videtur, “it seems”) is appropriate here because it is unclear to what extent the reference to Joachim is coming from Irenaeus’s (or his Latin translator’s) text of Matt 1:11 or from Jer 22:24’s “Jeconiah, son of Joachim” which Irenaeus proceeds to quote immediately thereafter.