Turquoise mosaic dragon and bronze bell in rich male burial at Erlitou, phase II, c. 18th century BC.

A really good historical source is coeval with the events it describes, or it may even form a part of those events, such as in the case of a land deed. It is written by a knowledgeable participant in the events, one who is not strongly politically biased or whose bias is at least known. And any statement in a good historical source is ideally corroborated by other independent good historical sources.

Now, in no part of the world is there any historical source older than the first proto-cuneiform inscriptions in Mesopotamia about 3300 BC. In most areas, the oldest sources are many thousands of years younger. But before the historical period in each area, there is usually a proto-historical era: one for which there are only few and bad sources, often quite extremely bad ones. Typically, proto-historical sources are written centuries after the fact by political propagandists, and there exists no corroborating historical evidence.

Proto-history offers a powerful lure to all students of the past: oh, how we all wish that we could somehow dig good historical knowledge out of those crap sources! Less than a century ago, certain Scandinavian scholars were still looking for the battlefield of Brávellir and trying to pair up the kings of the “Ynglingatal” with great grave barrows at Old Uppsala and elsewhere. All futile, a chasing after the wind. Those texts must be treated like fairytales, because they most likely are and there is no way of finding out if they aren’t.

Everywhere on earth, the proto-historical sources that appear to stretch the farthest inte the past are usually genealogies, often royal ones. “Lo, King Freddie reigned for 253 years and begat King Ronnie who smote the Fellatians and reigned for 346 years and begat King Reginald who reigned for 123 years and begat King Humpty” etc. At the head of the list is usually a god who acts as mythical ancestor of whoever is king at the time when the genealogy is compiled.

For a blatant example, some Anglo-Saxon king lists are headed by Odin and Julius Caesar. A Sumerian king list from about 2100 BC appears to reach more than 1900 years into the past, but better evidence shows that the kings on the list all lived within a 600-year interval. The compiler of the king list has gathered royal genealogies from several coeval Sumerian city states and stacked them on top of each other to get a really long list.

As historical scholarship has improved, one area after another on Earth has lost its proto-historical innocence and relegated the king lists to a position as sources on the royal propaganda of far later ages than they purport to treat. But this process isn’t complete yet. A good paper by Liu Li & Xu Hong in the current issue of Antiquity (available behind a pay wall) shows that China is finally beginning the necessary reevalutation. This is a painful process, as the current Chinese elite quite likes the idea of unbroken historical continuity way back into the Bronze Age. But it is encouraging to see that although Liu is an expat working in Australia, Xu is affiliated with the Chinese Academy of Social Sciences in Beijing. The English of their paper is excellent.

The first historical sources for China are brief inscriptions from the third quarter of the 2nd Millennium BC, found at palatial sites in the Yellow River valley. Some names of kings mentioned in oracle bone inscriptions from about 1300 BC onward are identified in much later historical texts as kings of the late Shang dynasty. The Shang gave way to the Zhou in about 1000 BC. But the royal genealogies used to reconstruct the early dynastic chronology date from after 300 BC, that is, they are at best products of the final decades of the Zhou dynasty.

Shang and early Zhou China are typical proto-historic periods, though many Chinese historians will tell you otherwise. Yet Chinese proto-history reaches even farther back. Before the Shang, say the late sources just mentioned, there was the Xia dynasty, a separate people that was conquered by the Shang around 1600 BC. This means that the archaeologists of the Yellow River Valley have been called upon to identify a noteable discontinuity in the material record of the period around this date. “Show us the Xia/Shang boundary!”, cry nationalistic historians.

Liu & Xu’s paper concentrates upon China’s earliest palatial site at Erlitou. It rose into prominence about 1900 BC from Neolithic village roots and flourished until about 1500 BC or later. This is inconvenient from the point of view of the royal genealogies, because they suggest that there should be a big break around 1600 BC. Liu & Xu demonstrate that no such break is actually to be found at Erlitou, despite many attempts by earlier contributors to the debate to massage the archaeological evidence. And Liu & Xu draw the obvious (yet far from uncontroversial) conclusion: if really bad historical sources are found to clash with the archaeological evidence, then this needn’t worry either archaeologists nor historians. The Xia dynasty — kings, dates and all — is simply a later fiction.