The original manuscript is not extant for the ending of the word armour(s); only the first three letters, arm, are extant in 𝓞. We cannot tell if the word read in the singular or plural—or if the base word was spelled armour or armor. The spelling issue is readily solved since Oliver Cowdery always spelled the word as armour in the manuscripts. Once he initially spelled the word as armor in 𝓞, but this he corrected to armour almost immediately (in Alma 46:13, listed below).
As far as the grammatical number is concerned, here in Alma 46:21 the printer’s manuscript has the plural armours. In current English we treat armor as a noncount noun, so the use of the plural seems unusual here in Alma 46:21. For this reason, both the LDS and RLDS texts have been edited to the singular armor (in 1874 for the RLDS text and in 1920 for the LDS text).
When we consider all instances of armor(s) in the text, we find that only here in Alma 46:21 is there any textual evidence for the plural armors; in the following list, there are four instances (including the plural one here in Alma 46:21) that refer to girding on armor, each one marked below with an asterisk:
More specifically, three of the four cases refer to girding armor “about one’s loins”; the one in Alma 3:5 does not, although earlier that passage refers to “a skin which was girded about their loins”. The exceptionality of armors for the earliest text in Alma 46:21 clearly suggests that armors could be an error for armor.
There are two basic possibilities for explaining why armors occurs in the earliest textual sources for Alma 46:21: (1) the original text read in the singular, as armor, but it was accidentally changed to armors, either when Oliver Cowdery took down Joseph Smith’s dictation or when Oliver copied the text from 𝓞 into 𝓟; or (2) the original text read in the plural, as armors, and this plural form was faithfully transmitted throughout the manuscripts and early editions.
It should first be noted that there is no specific manuscript evidence that Oliver Cowdery or any of the other scribes ever mixed up the number for armor. This finding provides some support for armors as the original reading in Alma 46:21. Nonetheless, there are so many examples of Oliver mixing up the grammatical number for other nouns in the text that we cannot rely on the lack of variation in number for nine instances of armor(s). Moreover, one could argue that armors in Alma 46:21 was an error induced by the plurals loins and garments in this sentence (“with their armors girded about their loins / rending their garments”). But such a proposed error did not occur in the three other instances (each marked above with an asterisk) where loins occurs close to armor.
Another factor worth considering is that Alma 46:21 is in that portion of the text where the 1830 sheet (the 22nd signature) was set from 𝓟 but proofed against 𝓞 (see the discussion under Alma 42:31). Both 𝓟 and the 1830 edition here read armours (or armors originally in the 1830 edition until emended to armours in an in-press change), which argues that 𝓞 read in the plural. Of course, it is always possible that the lack of an s in 𝓞 could have been missed in proofing. See, for instance, the discussion below regarding verse 22 where an original should in the manuscripts was mis-set as shall in the 1830 edition but never corrected.
Historically, in Early Modern English there is evidence that armor was frequently used in the plural when referring to the armor of more than one person (or animal). We have the following examples in the plural under armour, definition 2 (‘a suit of mail’), in the Oxford English Dictionary (here I retain the original accidentals):
Thus the plural armors is possible in Alma 46:21 since the reference is to more than one person. The plural reading is the earliest extant reading (in 𝓟); the critical text will therefore accept armors, despite its difficulty for modern English readers.
Summary: Restore the plural armors in Alma 46:21, the earliest extant reading (in 𝓟) as well as the reading of all the early editions; although a difficult reading, there is evidence from Early Modern English for such plural usage.