Tau in its oldest form consists of two straight lines of like thickness, the horizontal being bisected by the lower and vertical one. As early as in Cod. Sinaiticus the horizontal line is made thin, and strengthened on the left side only by a point or small knob (Nos. [3], [11]): thus we find it in Cod. Laud. of the Acts sometimes. In Cod. Alex. both ends are slightly pointed, in Codd. Ephraemi, Rossanensis, and others much more. In Cod. Bezae the horizontal is curved and flourished; in the late uncials the vertical is very thick, the horizontal fine, and the ends formed into heavy triangles (e.g. No. [16]).
Upsilon on the Rosetta stone and Herculanean rolls is like our Υ, all the strokes being of equal thickness and not running below the line: nor do they in Hyperides or in Codd. XZ and Augiensis, which have the upper lines neatly curved (Nos. [6], [9], [18], [38]). The right limb of many of the rest is sometimes, but not always curved; the vertical line in Codd. Vatican. and Sinaiticus drops slightly below the line; in Codd. A, Ephraemi, Cotton Genesis, Cureton's Homer, Laud. of the Acts and Rossanensis somewhat more; in others (as Codd. Bezae NR) considerably. In the subscription to St. Matthew's Gospel, which may be by a somewhat later hand, a horizontal line crosses the vertical a little below the curved lines in Cod. Rossanensis. In later uncials (Nos. [7], [36]) it becomes a long or awkward Y, or even degenerates into a long V (No. [16]); or, in copies written by Latin scribes, into Y reversed. We have described under iota the custom of placing dots, &c. over upsilon. But in Tischendorf's Leipzig II. (fragments from Numbers to Judges of the seventh or eighth century) upsilon receives two dots, iota only one. Once in Cod. Z (Matt. xxi. 5) and oftener in its Isaiah a convex semicircle, like a circumflex, stands over upsilon.
Phi is a remarkable letter. In most copies it is the largest in the alphabet, quite disproportionately large in Codd. ZL (Paris 62) and others, and to some extent in Codd. AR, Ephraemi, Rossanensis, and Claromont. The circle (which in the Cotton Genesis is sometimes still a lozenge, see above, p. [32], note 1), though large and in some copies even too broad (e.g. No. [18]), is usually in the line of the other letters, the vertical line being produced far upwards (Cod. Augiens. and Nos. [16], [41]), or downwards (No. [10]), or both (No. [36]). On the Rosetta stone the circle is very small and the straight line short.
Chi is a simple transverse cross (Χ) and never goes above or below the line. The limb that inclines from left to right is in the uncial form for the most part thick, the other thin (with final points according to the practice stated for epsilon), and this limb or both (as in Cod. Z) a little curved.
Psi is a rare but trying letter. Its oldest form resembled an English V with a straight line running up bisecting its interior angle. On the Rosetta stone it had already changed into its present form (Ψ), the curve being a small semicircle, the vertical rising above the other letters and falling a little below the line. In the Cotton Genesis psi is rather taller than the rest, but the vertical line does not rise above the level of the circle. In Codd. ANR and Rossanensis the under line is prolonged: in R the two limbs are straight lines making an angle of about 45° with the vertical, while oftentimes in Hyperides and Cod. Augiensis (Scrivener's photograph, ll. 18, 23) they curve downwards; the limbs in N and R being strongly (slightly in Rossanensis) pointed at the ends, and the bottom of the vertical bevelled as usual. In Cod. B of the Apocalypse, in Evan. OWdΞ, and even in Hyperides, the limbs (strongly pointed) fall into a straight line, and the figure becomes a large cross (No. [7]). In Evan. 66 the vertical is crossed above the semicircle by a minute horizontal line.
Omega took the form Ω, even when omicron and theta were square; thus it appears on the Rosetta stone, but in the Hyperides and Herculaneum rolls it is a single curve, much like the w of English writing, only that the central part is sometimes only a low double curve (No. [10], l. 6). In the Cotton Genesis, Codd. Vatican., Sinaiticus, Alex., Ephraemi, Bezae, Claromont., Nitriens., Rossanensis, there is little difference in shape, though sometimes Cod. Vatican. comes near the Herculanean rolls, and Cod. Alex, next to it: elsewhere their strokes (especially those in the centre) are fuller and more laboured. Yet in Cod. N it is often but a plain semicircle, bisected by a perpendicular radius, with the ends of the curve bent inwards (No. [14], l. 2). In the late uncials (Nos. [7], [16]) it almost degenerates into an ungraceful W, while in Cod. Augiensis (photograph, l. 18) the first limb is occasionally a complete circle.
These details might be indefinitely added to by references to other codices and monuments of antiquity, but we have employed most of the principal copies of the Greek Testament, and have indicated to the student the chief points to which his attention should be drawn. Three leading principles have perhaps been sufficiently established by the foregoing examples:
First, that the uncials used in writing differ from the capitals cut in stone by the curved shapes which the writing hand naturally adopts[34].
Secondly, that the upright uncials of square dimensions [pg 041] are more ancient than those which are narrow, oblong, or leaning[35].
Thirdly, that the simpler and less elaborate the style of writing, the more remote is its probable date.