Before the summer holidays, I made a few deeply satisfying, although geeky, discoveries in the course of my research. It's about gold bracteates. They're beautiful little things, not well known except to hardcore archaeology enthusiasts, found mostly in Denmark, southern Sweden and southern Norway. Made in the later part of the Migration Period (about AD 450 to 540), gold bracteates are little pendants bearing detailed mythological images, including the first recognisable scenes from the Scandinavian polytheistic mythology codified much later by Snorri Sturluson. Bracteate imagery borrows heavily from Roman medals of the 4th century, but is full of Scandinavian idiosyncrasies.
One of the most common motifs has a quadruped looking a bit like a horse, with a huge bodiless man's head sitting on its back. We call them "C bracteates". The man's lips often touches the horse's ear, and according to one interpretation the scene depicts Odinn in his role as healer of Balder's collapsed horse. This idea comes down to us through the Merseburg Incantations
Bracteates are very nice, but I didn't really look into them for their own sake. The reason had more to do about my general ideas of archaeological methodology, and there was a personal reason too.
One of the books that has had the strongest influence on my way of thinking about archaeology is Mats Malmer's 1963 collection of essays Metodproblem inom järnålderns konsthistoria
, "Problems in the Methodology of Iron Age Art-Historical Studies". A main point argued there is that no scholarly statement is worth anything unless the terms used are well defined and unambiguous -- for instance, when an archaeologist assigns source material to various types. Another important section analyses and establishes the theoretical background of typological seriation, one of the main tools used by archaeologists to order finds and sites in chronological phases. The book's central case study, where Malmer demonstrates how he thinks things should be done, concerns the chronology of the gold bracteates.
Ever since the 60s, many archaeologists have felt that while Malmer's ideas are sound in principle, and though he applied them successfully in other studies, still there's something wrong with his bracteate chronology. And since the late 70s, Danish scholar Morten Axboe has been studying the bracteates, with their chronology being a central line of inquiry. I've known since 1999 that Axboe's results aren't at all compatible with Malmer's, even though he uses an updated, computer-supported version of Malmer's methodology. And since then, I've been itching to find out why.
I know Malmer and Axboe, both are really nice guys, very supportive of younger colleagues such as myself, and I have the greatest respect for them as scholars. Malmer is in fact my number one archaeological hero. So I really needed to find out what was wrong.
Last year, Axboe finally published an extremely solid study of the bracteates' chronology. As soon as I'd read it, I started fiddling with the data. I tweaked Axboe's system a bit to make it work even better, and I juxtaposed it with Malmer's system.
What Malmer said in 1963 is basically that C bracteates can be divided into a) an early group, where the man has one of two particular hairstyles and the horse sometimes has a goatee beard; and b) a late group where a range of other hairstyles are seen, where the horse never has a beard, but where it sometimes throws a hind leg up over its rump. The beard and the thrown-up hind leg hardly ever occur together, and Malmer argued that the only likely explanation was chronology.
Axboe could process a far greater number of details thanks to computer supported multivariate statistics. He only recorded the details of the men's heads in order to be able to study bracteates without any animals as well. I collected data on the horses' beards and hind legs too, stuck them into Axboe's database and re-ran the stats.
What came out was really cool. Turns out that Malmer's two groups are both in fact very long-lived and contemporanous
. The reason that horse beards and certain hairstyles cluster together and avoid thrown-up hind legs and other hairstyles must be that the two groups of traits are depictions of two different men, each with his own individual horse. Combining their characteristics would apparently have been like drawing a Disney character with a duckbill and
mouse ears. The beard and the hind leg are what art historians call "iconographical attributes".
So my appreciation of both scholars is undiminished. Malmer's principles are indeed sound, he just chose the wrong material to apply them to. Gold bracteates are enormously complicated iconographical objects with lots of details, most of which are not relevant to chronology. And, from an archaeological point of view, they were made during a very short period of time, during which little dramatic change could take place.
Archaeology can usually place an object in the right century on stylistic grounds, but Malmer was trying to subdivide a group of finds that were all made during less than a century. Axboe, thanks to computer support and years of work, finally succeeded in dividing them into four phases.
Axboe, M. 2004. Die Goldbrakteaten der Völkerwanderungszeit. Herstellungsprobleme und Chronologie.
Ergänzungsbände zum Reallexikon der germanischen Altertumskunde 38. Berlin.
Malmer, M.P. 1963. Metodproblem inom järnålderns konsthistoria.
Acta archaeologica Lundensia, series in octavo 3. Department of Archaeology, University of Lund.
[More blog entries about archaeology, art, migrationperiod, Denmark; arkeologi, konst, folkvandringstiden, Danmark.]