Earlier I promised a thread on when most people converted to Islam. 125 years ago Thomas Arnold simply assumed that most Middle Easterners converted to Islam within 50 years of Muhammad's death. Most non-scholars still think Muhammad came, then boom, everyone's Muslim!
Fifty years later, Daniel Dennett (not the philosopher, his father) published a study which (inter alia) pointed out the lack of evidence for conversion in the Umayyad period, leading him to conclude that most people converted to Islam around 750, the Abbasid revolution.
In 1979, Richard Bulliet published Conversion to Islam in the Medieval Period: An Essay in Quantitative History, which pushed the timeline back even farther. He suggested that Iran became 90% or so Muslim by around 890 CE, but Iraq & Syria took about a century longer.
When I criticize Bulliet's argument, it is important to keep in mind that he sharpened the regional focus that Dennett had introduced, and that he was pushing later the timeline that scholars presumed before. On these points, my complaint is that he didn't go far enough!
There is a difference between historical arguments based on interpreting evidence and the numerical arguments advanced by Bulliet: when interpretive arguments fail, they are usually approximately correct, but when the wrong math is used, the true answer can be unrelated!
Outside of Iran, Bulliet's conversion curves were based on essentially two points: an assumed 3% of the population converted by 64 AH, and the calculated peak of "Muslim names" the moveable parameter establishing when he assumes 84% of the population had converted.
But as I pointed out elsewhere, at least 3 of the uniquely "Muslim names" were not uniquely Muslim
Furthermore, outside of Iran he is using essentially one primary source: al-Dhahabi's Kitab al-I`bar, and its abbreviation by Ibn al-`Imad.
If we take out the mistaken use of "Muslim names," his method would imply that, by the time these biographical dictionaries were authored, the population of each region was not only 100% Muslim, but also 100% ulama. That didn't happen by 1348, nor even 1679!
Imagine trying to do US demography from the ancestors of Christian seminary professors.
The odd thing is that despite these errors, and 42 years later, the book continues to be cited, as the last systematic attempt to grapple with when the population converted to Islam.
The problem is the book's math is wrong, so it tells us nothing. We simply don't know when most of the Middle East had converted to Islam (or other factors like immigration, different birth-rates).
I took a different tack, using geographies rather than biographies.
I learned that Islamization happened very locally, not at the scale of large regions, but individual villages or neighborhoods.
There was no large-scale "tipping-point," because 51% doesn't matter outside of a democracy. We probably couldn't distinguish 45% from 55%.
But in Syria there were still very few rural/village mosques around 1000 CE. In an agrarian society, the bulk of the population is rural. If the bulk of the population of Syria in 1000 CE had no mosques, it seems hard to believe the majority were Muslim, let alone 90%.
Iraq is less clear, but there I found evidence that the ecclesiastical hierarchy and monastic system was not significantly reduced by 1000 CE, which I would expect to see if 90% of the Christian population had converted to Islam and stopped their donations.
Egypt was long assumed to have become majority Muslim after the failure of the last Coptic revolt in the 830s, based on a remark by al-Maqrizi (d. 1412). But in 1973 Friedmann pointed out that this misreads al-Maqrizi's Arabic. It refers not to conversion, but victory!
So we must take seriously the possibility that Muslims were a demographic minority even after 1000 CE, even in areas that remained under Muslim rule. This different context may reshape our understanding of how religion functioned in society (to be continued...)