The traditional text, as reflected in translations like the AV and NKJV, reads "70." The modern critical text’s use of "72" is an example of an unnecessary change in the traditional text.
I. External Evidence:
The pivotal question is the inclusion or omission of one word, duo. Include the word, and it reads "72." Exclude the word, and it reads "70."
1. Greek manuscripts that include duo (Luke 10:1):
P75 B (Vaticanus) D (Bezae) 0181
In addition, this reading is supported by a few Old Latin manuscripts, the Sinaitic Syriac, the Curetonian Syriac, the Sahidic, and a single Bohairic manuscript. In the Church Fathers, it is found in some manuscripts of Origen (d. 254 AD) and in Adamantius (c. 300-350 AD).
2. Greek manuscripts that exclude duo (Luke 10:1):
Aleph (Sinaiticus) A C L W Theta Psi family 1 (1, 118, 131, 209, 1582) family 13 (13, 69, 124, 174, 230, 346, 543, 788, 826, 828, 983, 1689, 1709) and the Majority of extant manuscripts.
In addition, the traditional text is supported by the Syriac Peshitta, the Syriac Harclean, and the Bohairic. It is also supported by quotations in the Church Fathers including Irenaeus (2nd cen. AD); Clement (c. 95 AD); and Tertullian (c. 220 AD).
One will notice that the two manuscripts most prized by modern critical scholars are divided in their reading. Vaticanus supports the non-traditional reading and Sinaiticus the traditional reading. There are only four Greek manuscripts that support the non-traditional reading.
The external evidence appears overwhelmingly to support the traditional text of Scripture. Metzger, however, can somehow say, "The external evidence is almost evenly divided" (this and all other quotes below are from Bruce Metzger, ed. A Textual Commentary on the New Testament [UBS, Corrected ed., 1975]: pp. 150-51). The attestation to the traditional text is ancient and widespread.
II. Internal Evidence:
Metzger notes that "The factors that bear on the evaluation of internal evidence are singularly elusive." He adds that although the majority of the six man UBS committee decided to include duo, they enclosed the word in square brackets "to indicate a certain doubt that it has a right to stand there."
A minority report is attached from Kurt Aland. He notes that the examples of "70" in the OT is "overwhelming; there are always 70 souls in the house of Jacob, 70 elders, sons, priests and 70 years that are mentioned in chronological references to important events." Meanwhile, the number "72" appears just once in Numbers 31:38 in reference to the number of cattle set apart for sacrifice. On Aland’s reading, however, this is what makes it all the more "astonishing" that the "72" reading occurs at all in Luke 10:1, 17. He attributes the "70" reading to "ecclesiastical normalizing." Thus, he argues that the number in question "should be printed without square brackets."
Aland accepts "72" as the more difficult, and thus the preferred, reading. There are, however, several very plausible explanations as to how this reading might have developed. One common solution that is offered concerns the listing of 70 nations in the Hebrew text of Genesis 10 and 72 nations in the LXX text of the same passage.
The strong external evidence, along with a plausible explanation of the internal evidence, leads us to affirm the traditional reading. What Aland dismisses as "ecclesiastical normalizing" we see as divine preservation of the correct reading to maintain a reliable text for the church. The early believers eventually rejected the reading of "72" as spurious. There is no compelling reason to abandon the traditional reading of "70."
III. Survey of English translations:
What do the various English translations do with Luke 10:1, 17? As one would expect those that self-consciously follow the traditional text (AV, NKJV) read "70." It is surprising to find, however, that not all translations that follow the modern critical text read "72."
One translation that does follow the modern critical text is the NIV (1984). The NIV reads "seventy-two" in Luke 10:1 and adds in a footnote, "Some manuscripts seventy." Given the evidence, it would be more appropriate to have such a note read something like, "The vast majority of ancient Greek texts reads seventy, although four Greek texts read seventy-two."
Several modern versions choose to depart from the modern critical Greek text here and to retain the traditional reading. The NASB (1995), though generally following the modern critical text, here reads "seventy." Likewise, the Holman-Christian Standard Bible (2003) also reads "seventy" with a note stating, "Other mss read 72."
More curious are the translation choices in the RSV stream. The RSV (1952) retains the traditional reading of "seventy," adding in a footnote, "Other ancient authorities read seventy-two." The NRSV (1989) likewise preserves the traditional reading of "seventy" with the same footnote as the RSV. The ESV (2001), currently being promoted as a replacement for the NIV in the evangelical church, departs from the RSV and NRSV but duplicates the NIV by reading "seventy-two" with an identical accompanying note adding, "Some manuscripts seventy." The same critique applied to the NIV note above is appropriate to the ESV note here.
It is clear that even many scholars and translators who accept the modern eclectic text of the New Testament, realize that the weight of the evidence for departure from the traditional reading of Luke 10:1, 17 is suspect. The NIV and ESV, however, abandon the traditional reading headlong. Oddly enough, the "evangelical" translations are the ones that promote innovation, while the "liberal" ones (e.g., NRSV) perpetuate the traditional reading.