In a previous post I gave some general examples of how spelling and pronunciation can drift between languages: the d in Muhammad is spelt in some early sources with an emphatic t (teth) or regular t (taw), so I wanted to show that such variant spellings are to be expected and needn’t cause alarm. Since then, two readers have been kind enough to explain exactly how these spellings (most likely) came about.
In Syriac the plosive consonants b p g k d and t when following a vowel may soften to the fricatives v f gh kh dh and th. This can cause trouble for foreign words: a Syriac reader would instinctively soften the d in Muhammad to dh (the ‘th’ sound in ‘this’, but not in ‘thin’). It’s plausible that scribes replaced the letter d with an emphatic t (teth), which does not soften, in order to best approximate the hard d of Muhammad. This is a smart solution, but entirely optional, and another early Syriac text (Hoyland, Seeing Islam, 116-7; Palmer, West-Syrian Chronicles, 1-4) prefers to stick with d. Thanks to the linguist Yusuf Gürsey for pointing this out.
The Pahlavi case, on the Arab-Sasanian coins, would seem to result from a different process. Where in Old Iranic a vowel was followed by t, in Pahlavi it had softened to d. But written Pahlavi remained archaic, so postvocalic d was still written as t. On the coins this is written in an Aramaic script with a taw. By analogy, the d in Muhammad – which follows a vowel – was also spelt with a t, as though it were a native Iranic word. Thanks to François du Blois for this.
Of course we don’t need these explanations in order to rubbish Robert Spencer’s claim that Mhmt is other than Muhammad; but it’s very satisfying to see how these aberrant spellings came about.