move corruptions inserted by their predecessors, for if the good rhymes of A and B had belonged to the original text, what possible motive could any one have had for turning them into the bad rhymes of C? Copyists, as a class, dislike anomalies and get rid of them if they can; they resemble the pet bear that beat off flies from the face of its sleeping master with a stone. Moreover, the interpolated verses composed by Persians are technically excellent. The ancient copyists as textual critics. The Turkish scribes, it is true, write verse which is often execrable both in form and metre (see the Búláq edition passim), but even they rarely indulge in rhymes of the sort which Jaláluʾddín permitted himself to use. It might be conjectured that the poem was revised by some of the author's friends after his death, when uncorrected copies were already in circulation; but this hypothesis would not account for the gradual and incomplete process of emendation exhibited in the MSS. Probably the first critical editors of the Mathnawí were the copyists themselves. We have seen how unfaithfully they dealt with it in one department, where the traces of their handiwork happen to be conspicuous; and we can study their habits and psychology in other fields which are less exposed to observation. Here are a few examples chosen at random: At I ٥٨٢‎ AC read B has the correction made by some one who took offence at the construction of طالب شود‎ with تان‎, and reads in the second hemistich

Archaic words and forms usually disappear from the later MSS. Thus at I ٨٥٥‎ C has می گسست‎, for which AB write ,می شست‎ while at I ٣٦٤٨‎ ABC have وبشكلد‎, and L بگسلد‎. At I ١٠٢٩‎ AB change شیرگیری سازی‎, the reading of C, into مکر و شیراندازی‎.

At I ٢۱٦۹‎ C has غير آن پيرو نبود‎ for which AB substitute .آن پیر او ندیدغير‎. The old forms آفسانه‎ with alif-i mamdúda (I