B reads
گر بگیرم برکت دندان مار . تاش از سر گرفتن نبود ضرار
At I ٣١١٥ C has
گفت روبه صد سپاس آن شیر را . کز پس آن گرگ را خواند او را
Here the rhyme is faulty, because a nímfatḥa is made to correspond to a fatḥa. AB therefore eliminate it and write
آن گرگ خواندروبه آن دم بر زبان صد شکر راند . که مرا شیر از پس
In some cases the fault is not removed but is partially remedied. At I ٢٣٤. C reads
با بزخم من رگ جانت برد . یا ترا چون من بزندانی برد
A amends this by writing in the second hemistich
با ترا چون من بزندانت برد
thus introducing a type of rhyme واپس تر شود and بر رود (instead of واپستر رود), which is not uncommon in the Mathnawí, but impairing the syntax, as l is now superfluous. It remained for B to restore the syntax and write
یا که همچون من بزندانت برد
Those who desire further evidence will find plenty of material in the critical notes. I think, however, that the instances which have been given are enough to establish what I undertook to prove, namely,
(1) that in many places the text of the Mathnawí was altered by early copyists, not capriciously, but with a definite object and in a particular way.
(2) that of the MSS. used for Book I C is the nearest to the original, and that A comes next in order of authenticity. It is to be noted that L, a later and inferior MS., has frequently retained readings of C which do not occur in A or B.
The inexact prosody of the poem is quite in keeping with its free, unelaborate, and generally colloquial style. No one is likely to argue that the copyists represented by A and B sought to re-
