Classifying the manuscripts of the Pericope Adulterae into families required many steps. The first step was to give each continuous text manuscript a preliminary assignment to one of Von Soden's seven families, namely, μ¹, μ², μ³, μ⁴, μ⁵, μ⁶, and μ⁷, or to mark them as ‘mixed’ or ‘unclassified’. This required establishing a preliminary pattern of readings that aligned with each family across multiple variant units. I established variant units by recording the differences between Robinson and Pierpont's 2018 Greek text and seven editions of the Greek New Testament, namely, Antoniades (1904 edition), Hodges-Farstad (2^nd edition), Nestle-Aland (28^th edition), Pickering (3^rd edition), Scrivener (1894 edition), Stephanus (3^rd edition), and Westcott-Hort. I also recorded the alternate Byzantine readings listed in the margin of Robinson and Pierpont's text. This process yielded thirty-nine variant units, which I then used to establish a preliminary pattern of readings that aligned with Von Soden's seven families. For μ⁵ and μ⁶, I followed the μ⁵ and μ⁶ readings listed in Robinson and Pierpont's text and marked a family as split for any variant for which Robinson and Pierpont list an alternate Byzantine reading for that family. For μ⁷, I followed the main Family 35 text of Pickering's third edition and marked the family as split when Pickering indicated in his footnotes that Family 35 was split.

Using Robinson's collation of the Pericope Adulterae and Von Soden's lists of manuscripts for μ¹, μ², μ³, and μ⁴, I created a pattern of readings for those four families. A particular variant was included in the diagnostic pattern of readings when it had at least twice as much manuscript support within a family as the variant with the next highest level of support within the family. For example, if variant 1 was supported by 13 𝔐⁴ manuscripts, and variant 2 was supported by 6 𝔐⁴ manuscripts, variant 1 was included in the diagnostic pattern of readings for 𝔐⁴. However, if variant 1 was supported by 12 𝔐⁴ manuscripts, and variant 2 was supported by 7 𝔐⁴ manuscripts, variant 1 was not included in the diagnostic pattern of readings for 𝔐⁴. I followed a similar methodology to determine the diagnostic pattern of readings for 𝔐⁵, 𝔐⁶, and 𝔐⁷.

Using the resultant pattern of diagnostic readings, I determined the percentage of alignment each manuscript had with each family. If a manuscript had at least 90% alignment with a particular family, and if its percentage of alignment was at least eight percentage points higher than the family of second closest alignment, the manuscript was assigned to that particular family. For example, if a particular manuscript had 93% alignment with 𝔐⁴, 84% alignment with 𝔐³, and lesser alignment with the other families, it was included in the 𝔐⁴ family. However, if a manuscript had 93% alignment with 𝔐⁶ and 86% alignment with 𝔐⁷, it was not included in either family because of its mixed nature. Similarly if a manuscript had 73% alignment with 𝔐⁵, 57% alignment with 𝔐⁶, and lesser alignment with the other families, it was not assigned to any family because it did not meet the 80% alignment threshold. This process resulted in a list of manuscripts for each family along with lists of mixed and unclassified manuscripts.

Using this new list of manuscripts for each family, I adjusted the diagnostic profile of readings for each family. I then repeated the process of calculating alignment with each family and redetermining the list of manuscripts for each family. After a number of passes through the data, the list of manuscripts for each family had stabilized. I followed similar procedures to determine previously unidentified families among the continuous text manuscripts and also among the lectionary manuscripts, which were treated separately from the continuous text manuscripts. In the process, I expanded the variant units under consideration for all families to a total of forty-nine. After my final pass through all the data, I lowered the alignment threshold to 80% to determine a final list of manuscripts for each family, while maintaining the standard that the percentage of alignment for a family had to be at least eight percentage points higher than the family of second closest alignment. I also identified some smaller groups of closely related manuscripts that I chose not to identify as families due mostly to their size.

The final lists of manuscripts are presented below. Tier 1 manuscripts have at least a 90% alignment with their family, while Tier 2 manuscripts have between an 80% and 90% alignment with their family. The readings of this edition have been established on the basis of Tier 1 manuscripts only. Tier 2 manuscripts are listed for academic purposes. A reading was assigned to a family if it had at least twice as much manuscript support within the family as the reading with the next highest level of support within the family. If a reading did not have at least twice as much manuscript support within the family as the reading with the next highest level of support within the family, the family was considered to be split.

To avoid presenting an overly complicated apparatus within the footnotes of this edition, only Von Soden's seven families are listed along with two newly documented families of lectionary manuscripts. All the variant readings for the editions of the Greek New Testament I compared are tied to at least one of those families. However, the other newly documented families are listed in full in an apparatus below.