Reconstructing the Historical Muhammad
Based on this, some historians have proposed significant differences between the Muhammad of history and that of Muslim historical memory and devotion. Patricia Crone, for example, has argued that he lived not in Mecca and Medina, but in the far northwest of Arabia near the Byzantine frontier, and supports the conclusions of another revisionist historian that his opponents were not polytheists, but rather followed a monotheism which Muhammad and his followers found flawed. While I have not read the debate over the second proposition in perhaps a decade, I reject her geographic relocation. This is partly because no one has to my mind proposed a convincing reason to go back and relocate him to the Hejaz. Crone says in the linked article that later Muslims wanted to show he had no Jewish or Christian influence, but contact with Jews and Christians is nonetheless found throughout the traditional Muslim accounts of his life.
One benefit historians of early Islam do have is a form of source citation called the isnad. This is a chain of authorities through which a given source learned of the event in question. This does not solve all of our problems with the sources, by any means. For one thing, the standard of using isnads developed only gradually. Much as Christians made up accounts of Jesus and the apostles, so did Muslims, and for a variety or reasons both savory and unsavory. They then, of course, forged the chain of authorities, as well. The huge multi-volume sets of canonical hadith one sees on library shelves are only a small part of what once circulated, a part which was declared authentic by a few Muslim scholars (two in particular) during the 9th century and gradually accepted as such in the centuries following. Most modern historians believe the compilers of these collections did not go nearly far enough in weeding out forgeries; some will even assert that there are no authentic hadith whatsoever.
The hadith corpus is actually different, however, than prophetic biography. The former has served primarily to build up the range of Islamic law through accounts of how people were to perform rituals or how Muhammad responded to questions and cases. The biographical tradition is more concerned with events that served as foundational to the community: in the idea of a prophetic community of which later Muslims were heir, as propaganda for or against later political leaders based on the conduct of their ancestors, and as the key to interpreting the Qur'an, which was revealed at particular times when Muhammad sought out divine wisdom in general or guidance in particular situations. (I should note, though, that which revelations go with which occasions is often a matter of hadith.)
In an essay in this book, Andreas Gorke discussed several ways historians working primarily with the prophetic biographies are approaching their sources. One is analogous to the "criterion of embarrassment" used by scholars of the historical Jesus. The idea goes that later Muslims would not invent something that put Muhammad in a bad light. Gorke identifies a problem with using this approach: You tend to wind up assuming that everything that makes Muhammad look bad is true and everything that makes him look good is later apologetics. I'd go even further and say that we have no secure grounds on which to say what would and would not have been made up across the long decades people were making things up. An interesting example of the problems with thinking you can comes from the story that the First Crusade massacred the Muslims in Jerusalem. Historians today reject it, but the account is not found in Muslim sources. It comes from a Christian source (I forget which) who thought he was idealizing the Crusaders by making them sound as much like the Maccabees as possible. We know this because we can easily compare the Crusades source with the text of Maccabees in Catholic/Orthodox Bibles and see the deliberate parallels, but for the century after Muhammad, we don't have that.
Another method is to look for archaic language, an example of which I described here. Qur'anic Arabic is often extremely obscure, and even the 9th and 10th century commentators often admitted they were not sure what some of its words meant. The Constitution of Medina is also regarded as authentic mostly on those grounds. There appear to be some authentic letters of Muhammad to people in northwestern Arabia (perhaps supporting Crone's views), though certainly not all the letters attributed to him to people all over the Middle East. Finally, many but not all scholars accept as authentic some letters which a man named Urwa b. al-Zubayr wrote to the Umayyad caliph Abd al-Malik, who came to power in the Muslim civil war of the 680's and reigned until 705. More on these in a moment.
The other way combines what Gorke actually describes as two. One is to look at different versions of accounts about Muhammad which have a common link. For example, Ibn Ishaq's biography mentioned above is found in recensions by both Ibn Hisham and al-Tabari. There are some differences, but the idea is that where they match is probably authentic material from Ibn Ishaq. From this you could then compile a bunch of Ibn Ishaq material, albeit understanding some of the material you rejected as being only in one later source might be authentic, or that where later sources contradict each other someone might have what Ibn Ishaq actually said, or that he might even have changed his mind and the later sources reflect differences between points of his life.
Based on this method, Gorke is one of several scholars who has set about trying to reconstruct the information about Muhammad that was passed on by Urwa b. al-Zubayr. Who was Urwa b. al-Zubayr? He was the prophet Muhammad's nephew, the son of a prominent Companion of the Prophet, grandson of the first caliph Abu Bakr, and nephew of Aisha, a daughter of Abu Bakr and Muhammad's youngest wife. Although himself born about 15-20 years after Muhammad died, he obviously had good sources, and himself usually claims to have gotten his information about Muhammad from Aisha. The idea is that the Urwa b. al-Zubayr corpus can serve as useful bedrock for reconstructing the Muhammad of history.
This method gets criticized. For one thing, the isnads involving Urwa might be forged. Urwa may have handed down incorrect information, either deliberately for personal or political reasons or accidentally. Even if he reliably transmitted from Aisha and others who knew Muhammad, human memory is often unreliable, seeing early events through the prism of later developments and moods. Nonetheless, it strikes me as the most promising direction currently undertaken in the field. Tomorrow, I'll post about the conclusions these scholars have drawn.
UPDATE: I wound up writing three posts on this: I, II, and III.