Here is the Classic Online Encyclopedia's entry, based on the Encyclopedia Brittannica (1911):
"...Von Soden introduces, besides a new notation of MSS. (see Bible, N.T. MSS. and versions), a new theory of textual history. He thinks that in the 4th century there were in existence three recensions of the text, which he distinguishes as K, H and I, with the following characteristics and attestations.
K corresponds roughly to Westcott and Hort's Syrian Antiochian text [= Byzantine text]; it was probably made by Lucian in the 4th century. This was in the end the most popular form of text, and is found in a more or less degenerate state in all late MSS. The purest representatives are 61(52), 75 (V), 92, (461), 94, 1027 (5), 1126 (476 = scrivener's k) ε179 (661). Later recensions of K are called K' and Kr , and there are also others of less importance which represent the combination of K with other texts.
H represents Westcott and Hort's Neutral and Alexandrian texts between which von Soden does not distinguish.
It is found in eleven MSS. in varying degrees of purity: δ1(B), δ2 (א), δ3 (C), δ6 (Ψ), 8 8 ε26 (Z), ε6 ε1026 (892), 6 37 1 (1241) and 376 (579). Between these MSS. there is no very intimate connexion except between δ1 and δ2 (B and א) which represent a common original (δ1-δ2). B is the best representative of H, but it has been contaminated by the Egyptian versions, and sometimes by the K and I texts and by Origen, though not to any great extent.
The other H MSS. are none of them equal in value to the two great uncials. They have all been influenced by K, I, and by the text of parallel passages, to a greater extent than δ1-δ2, or than either of the two witnesses to δ1-δ2, but some of them have less Egyptian corruption.
The origin of the H text must be regarded as unquestionably Egyptian, in view of the fact that it was used by all the Egyptian Church writers after the end of the 3rd century, and von Soden adopts the well-known hypothesis, first made popular by Bousset, that it represents the recension of Hesychius.
I does not quite correspond to anything in Westcott and Hort's system, but has points of contact with their "Western" text. It is found in a series of subgroups of MSS. known as Hr, J, Ia , and others of less importance (about eleven subgroups are suggested). Of these Hr is a family [Family 1] containing Codex 1 and its allies (6254, ε34 6, 6 457, 5 467, &c.), ε288 (22) and some allied MSS. ε203 (872), ε183 and 1131; J is the well-known Ferrar group; [family 13] and Ia contains δ5 (D), 93 (565), (700), 050 and some others. It is necessary to note that von Soden is able to place D in this group because he regards it as owing many of its most remarkable readings to contamination with the Latin version. I is, according to von Soden, a Palestinian recension connected with Eusebius, Pamphilus and Origen.
After establishing the text of I, H and K, von Soden reconstructs an hypothetical text, I-H-K, which he believes to have been their ancestor. He then tries to show that this text was known to all the writers of the 3rd and 2nd centuries, but has naturally to account for the fact that the quotations of these writers and the text of the early versions often diverge from it. The explanation that he offers is that the Diatessaron of Tatian was widely used and corrupted all extant texts, so that the Old Syriac, the Old Latin, the quotations of Irenaeus, Clement, Tertullian and others may be regarded as various combinations of the Tatianic text and I-H-K. Finally, he tries to show that the Tatianic text is itself in the main merely a corrupt form of I-H-K altered in order to suit the necessities of Tatian's plan.
For criticism of this important theory up to 1909 see Nestle's Pinfiihrung in das griechische neue Testament, pp. 274-278 (3rd ed., Göttingen, 1909), and K. Lake's Professor H. von Soden's Treatment of the Text of the Gospels, Edinburgh, 1908). (K. L.)
I had to clean up their MS citations, but this may not be 100%.
From a review of The Textual Tradition of the Gospels: Family 1 in Matthew. By Amy S. Anderson. (New Testament Tools and Studies, 32.) Leiden: Brill, 2004
"Family 1 of the Gospels was edited by Kirsopp Lake from four or five manuscripts in 1902 and has been accepted as an entity and quoted as such ever since. Hermann Freiherr von Soden, in his work begun in 1902, identified a somewhat enlarged group of witnesses, including those used by Lake, giving it the sigla Hr in the analysis volumes, changed to Iη in the text volume. Amongst those enlarging the group was one of the many manuscripts brought to light by his team of young scholars, in this instance from the Athos monastery of Vatopedi, bearing the call number 949. To this was assigned the serial (now GA= Gregory–Aland) number of 1582. It is not known who the discoverer was; indeed Anderson's form of words does not mention von Soden's enterprise as the locus of discovery at all. The English scholar B. H. Streeter spoke highly of 1582 in his book The Four Gospels, drawing attention to significant readings and important critical notes in the margin. After the publication of von Soden's work, Silva Lake, in her revision of Kirsopp Lake's book The Text of the New Testament (1928), included 1582 in her account of Family 1.