The context is a census of the Levites: "Number the children of Levi by their families…. So Moses numbered them according to the word of the LORD, as he was commanded." (3:15-16).
The number of males one month and older is given for each of the three Levitical clans: the Gershonites: 7,500 (v. 22); the Kohathites: 8,600 (v. 28); the Merarites: 6,200 (v. 34). Then the total is given in Numbers 3:39: "All who were numbered of the Levites, whom Moses and Aaron numbered at the commandment of the LORD, by their families, all the males from a month old and above, were twenty-two thousand." And here is the difficulty. If we add up the previous figures (7,500 plus 8,600 plus 6,200) we arrive at 22,300. Why does this number not equal the 22,000 figure in Numbers 3:39? Certainly the number is not insignificant. The Levites will take the place of the firstborn of Israel (v. 12), but when the number of the firstborn is reckoned, it is 22, 273 (v. 43). This is 273 more than the 22,000 Levites of 3:39, so the Lord allows the Israelites to pay a redemption price of 5 shekels for each of the 273 surplus firstborns (see vv. 46-51).
Back to our difficulty. How do we reconcile the totals of 22,300 (vv. 22, 28, 34) and 22,000 (v. 39)?
I. Here are some explanations:
1. Textual variant in v. 28.
Some point to a textual variant in some texts of the Septuagint. The NKJV note in v. 28 before the "six" reads, "Some LXX mss three." Indeed, the Hebrew word for "six" is sh-sh and the word for "three" is sh-l-sh. If the number of Kohathites was 8,300 rather than 8,600, then the total would be 22,000 equaling the number in 3:39.
John MacArthur takes this view in his Study Bible note on v. 28: "The Kohathites probably numbered 8,300 males (See the marginal note on 3:28; the addition of one Hebrew letter changes the ‘six’ to a ‘three.’ This letter was dropped very early in the copying of the text)" (p. 201).
2. The use of round numbers.
Another explanation is that the author of Numbers is using round numbers. The 7,500, 8,600, and 6,200 are all even numbers. Could it be that the 22,000 in 3:39 is a rounding off of the 22,300? Are we seeking more precision than the text demands?
The number of 22,300 Levites is maintained, but 300 are excluded as firstborns themselves.
This is apparently the approach taken by the Talmud. J. J. Owens in his Numbers commentary writes: "Since there were firstborn themselves among the Levites, they could not release a corresponding 300 firstborn of the non-Levites" (p. 93). Matthew Henry also takes this view: "but it is supposed that the 300 which were struck off from the account when the exchange was to be made were the first-born of the Levites themselves, born since their coming out of Egypt, which could not be put into the exchange, because they were already sanctified to God."
II. Evaluation of explanations:
The textual variant explanation must be rejected. The LXX should not be used here to correct the Hebrew which consistently has the number of Kohathites as 8,600 and not 8,300. We must remember that the scribes who maintained the received text saw no conflict in preserving the figures as they are present in the current text. They knew math, but they saw no contradiction.
The round number explanation must also be rejected. The key consideration here is the exactness given in the discussion of the redemption of the firstborn in Israel. Their number is the very clearly not-round 22,273, and the 273 surplus is clearly given.
The Talmud explanation is more tenable, but this explanation is also not clearly stated within the text itself.
We must conclude that for some unexplained reason, 300 of the 22,300 Levites were excluded from consideration in the redemption of the 22,273 firstborn Israelites. One possible explanation is that these were the firstborn of the Levites themselves.
Does this challenge our understanding of Scripture’s infallibility? No. It simply reveals that though Scripture itself is always clear, our ability to understand it is not always sufficient.