INSCRIPTIONS XIX we have to supply omissions rather than to correct error. The chances of error (cp. for examples C. I. A. 151, p. 72: 398, 419, 483) except in mere spelling are comparatively small. There are no recensions of the text ; no glosses which have crept in from the margin, or inferences from the words of scholiasts that the reading may have been originally different. Far greater pains and time are neces- sarily taken in engraving than in writing ; and, speaking generally, inscriptions are at first hand and there is no further risk from copying. The greater danger is from the unskilfulness or ignorance of the modern cop3ast, but the original is generally in existence, and the error can be corrected. Whereas MSS. have been written and rewritten many times, at each rewriting contracting some degree of inaccuracy, and changing to a certain extent their modes of spelling and forms of grammar in successive generations. Pen and ink are more pliable implements than the chisel, and the writer takes greater liberties than the engraver in the form and size of the letters. But inscriptions are monumental, and the words and letters in them have a fixed character ; or, at any rate, only change with well- known changes in the alphabet. Almost invariably' in inscriptions of the fifth century' each letter occupies the same space, and in supplying lacunae, however large, we can measure with a compass the number of letters required. Wherever the graver has been skilful the symmetry is perfect, and a straight line may be drawn horizontally, verticall}', diagonally through the centre of the letters. But in some cases the miscalculation of space has led to the crowding of the latter part of the inscrip- tion : and there are other examples (cp. C. I. A. Suppl. i, 61 a, 71 ; Newton and Hicks, p. 61, 85) in which the lines are not written accurately aroLxq^ov. Many of the later inscriptions difter from the earlier ones as much as the fairest copperplate from the first rude attempts of an
- See Meisterhans, § 4.
