When dealing with material culture, there is one big difference when it comes to analysis. While in the previous example we start from the individuals and their relationships, in archaeological analysis we start from the material remains left from these relationships. As a result, we are taking a step back.
Let’s return to the previous example. Imagine archaeologists have found a number of items that belonged to these people. In one house (e.g. Anna’s), archaeologists have unearthed the remains of a recognizable pen and a sports bag. In another house (e.g. Maria’s), archaeologists have found the remains of a similar sports bag, as well as a mailbox key. Then, finally, in yet another house (e.g. Luke’s), archaeologists have unearthed another pen similar to the one found in Anna’s house, as well as a mailbox key, similar to what was found in Maria’s house. The latter two dwellings (Maria’s and Luke’s houses) are also found close together, let’s say they are part of the remains of an apartment building.
You can suppose that these different types of objects – the pens, the sports bags, and the mailbox keys – all look alike because the aforementioned people were using the same gym, worked at the same office, or lived in the same building. But this is all circumstantial: the similarity could be due to other factors, i.e. two different gyms using the same brand for the sports bags they give out to clients. If you are lucky, you could find some writing or a logo or a trademark on the objects, showing that the objects came from the same gym or the same office.
Nevertheless, even in that case you cannot rule out other possibilities: i.e. a gym having more premises, or the same company having multiple offices. In the case of the mailbox key, the fact that they were found together suggests that these people lived in the same building.
Thus, when dealing with material culture the reconstruction of the relationships and the use of network analysis is more indirect, and direct relationships cannot be postulated a priori without further evidence. At most, we can talk about a range of similarities, which suggest a similar environment: Anna and Maria probably went to the same gym, but they may well have gone to different classes or visited different locations owned and operated by the same chain. It doesn’t automatically follow that they would have known each other.
When written texts are available, the situation is not always made easier. For example, some people could be referred to only by their title or a nickname, thus we don’t know precisely who they are. There could also be homonyms, or other factors could come into play. Going back to the previous example, imagine that Maria uses her original surname on her gym card, but on the key card she uses the name of her spouse. We could not be sure that we are talking about the same Maria, unless we have other proof that she is the same person (e.g. a marriage certificate).
Network analysis is not some kind of magic mixer where we put fruit in and always get the same juice: the juice we get depends on the fruit that we use. In the same way, the graphs that are generated as part of network analysis depend a lot on the data we decide to include and on other small – but not trivial! – decisions.
I will write about these issues in future articles, and how it is possible to make the most out of network analysis, even when the data are more limited, as is the case with the Second Intermediate Period.