When dealing with the problem of determining which text is original there are two categories of evidence: external and internal. External evidence deals with the manuscripts in which the text appears. Internal evidence looks at the text and attempts to determine why the error was made and how the error relates to the text as a whole.
External Evidence regarding Romans 8:28
Although the external evidence numerically affirms the omission of ο θεος it becomes immediately apparent that some important evidence includes these words (P46 A B 81 sa (eth) Origengr2/5). These are important witnesses, including Origen, and are not easily dismissed which is why this decision is such a difficult one. P46 is the earliest manuscript of the Pauline letters and dates to about 200. It includes all his letters with the exception of 2 Thessalonians, Philemon, and the Pastorals. In Romans it includes 5:17-6:14; 8:15-15:9; 15:11-16:27. Traditionally it has been noted that P46 contains a “free” text. In other words it predates the development of text types. This categorisation is now questionable but the value of this papyrus is not. This is corroborated by Gordon Fee who quotes Gunther Zuntz at length:
The excellent quality of the text represented by our oldest manuscripts, P46, stands out again. As so often before, we must be careful to distinguish between the very poor work of the scribe who penned it and the basic text which he so poorly rendered. P46 abounds with scribal blunders, omissions, and also late additions. In some of these the scribe anticipated the errors of later copyists, and in some other instances he shares an older error; but the vast majority are his own uncontested property. Once they have been discarded, there remains a text of outstanding (though not absolute) purity.
Manuscripts A and B are also of value. The Alands make the statement that “in the gospel A transmits an exemplar with a rather poor text, but beginning with Acts its quality changes remarkably; in Acts it is comparable to B and א, while in Revelation it is superior toא and even P47” (303; 50). Codex Alexandrinus (A) is from the fifth century whereas Vaticanus (B) is from the fourth. The Alands rate B “by far the most significant of the uncials” (109). Both A and B are category I manuscripts (high proportion of early text). Minuscule 81 is also worth a mention as it rates at least category II (manuscript of a special quality; considerable proportion of early text, but which is marked by alien influences; Aland, 335). In this case there can be no question that the reading is an early one.
The most important manuscript in support of the text isא Codex Sinaiticus from the fourth century. It is rated category I, but was overrated by Tischendorf “and it is distinctly inferior to B, together with which (and P75) it represents the Alexandrian text” (107). C is a category II manuscript from the fifth century. Dp Codex Claromontanus which dates to the sixth century omits Romans 1:1-7 and 1:27-30 has been added by a later hand. Ironically, Dp does not belong to category IV, the “D” text but rather to category II (Dp 06*) and category III (Dp 06c). Fp and Gp are both Greek-Latin diglots from the ninth century. F is category II and G is III. Ψ is a category III manuscript from the eighth/ninth century. The minuscules are a range of qualities with the majority of them included in category III. For the sake of space I will list them by category along with the uncials:
Category I:א 33. 1175. 1739.
Category II: C D F 256. 1506. 1881. 1962. 2127. 2464.
Category III: G Ψ 6. 104. 263. 424c. 436. 459. 614. 1241. 1319. 1573. 1852. 1912. 2200.
Category V: K L P 424*.
What is particularly interesting about all of this is that although the superior manuscripts include the text ο θεος the omission has widespread geographical and church support along with the versions. It should also be noted that the Alexandrian text type is not unanimous asא Ψ 1739. 33 and 104 support the omission. Clement of Alexandria (ca. 180 – ca. 215) also appears to have a text that exclude ο θεος. The Western text type (DFG latt) also supports the omission as does the Byzantine, which might be expected given this. This makes it exceedingly difficult to come to a conclusion based on external consideration although overall it favours omission. In the next post we will turn our attention to internal evidence which will hopefully be more conclusive.
 Kurt Aland and Barbara Aland. The Text of the New Testament. 2nd ed. Trans. Erroll F. Rhodes. Grand Rapids: Eerdmans, 1989. 93.
 Gunther Zuntz. The Text of the Epistles: A Disquisition upon the Corpus Paulinum. Schweich Lectures, 1946. London: British Academy, 1953. 212-213. Quoted in Gordon Fee. “Rigorous or Reasoned Eclecticism – Which?” In Studies in the Theory and Method of New Testament Textual Criticism. Eldon Jay Epp and Gordon D. Fee. Grand Rapids: Eerdmans, 1993. 128.
 א Ψ 1739 always represent a key expression of the Alexandrian text type.
 The Stromata 4.7. I have only found it in English translation but the quote appears in a block of quotations from Romans: 8:7-8,10,13,17-18,28-30 and reads And we know that all things work together for them that love God. The Ante-Nicene Fathers. Vol. 2. Ed. Alexander Roberts and James Donaldson. Grand Rapids: Eerdmans, 1985. 417.