Many different processes can influence the position of events on the landscape,
individually or operating together. In discussing these processes, a useful distinction to make is between first and second order effects. The critical difference between the two is that first order effects operate such that each event in located independently of other events, while with second order events the location of one event is influenced by the locations of other events; essentially there is an interaction between events which influence their locations. One form of first order effect occurs where event location is influenced by specific topographic features on the landscape. Are events randomly
distributed over the landscape or are certain subsets of the landscape preferred? How do soil types, forest cover, mountain passes, rivers etc. impact the location of events? One example can be found in the Ontario Iroquoian Tradition, where Early Ontario Iroquoian sites seem to be found on sandy soils while Late Ontario Iroquoian sites are found on clay soils (Pearce 1996). Similarly, large expanses of swampy land might be almost devoid of site location while higher ground around the swamp would seem to be a preferred
location. Another first order effect, which can impact the distribution of events, is the presence of navigable waterways used as transport routes. Hodder and Orton (1976) illustrate an analysis where the patterning could not be reasonably understood until the occurrence of rivers as transportation routes was considered. The critical hallmark of first order effects is that the apparent variation can be largely explained by reference to the landscape over which the events are distributed. Another useful example of the concept of first order effects occurs in epidemiology, where one could be examining the
occurrence of instances of a specific disease. One could easily plot the incidences on a map and look for clusters, but a problem occurs when the at-risk population is not randomly distributed over the landscape but tends to clump together in cities, towns and villages. The apparent clusters from the plotting of the disease might just be reflecting the distribution of the at-risk population. In this case, the distribution of the population over the landscape would be considered as a first order effect. Analogous situations present themselves frequently in archaeology. Examples discussed in later chapters are the distribution of artifacts over a site or graves in a cemetery.
In contrast, second order effects are characterized by the interaction between two events. Essentially the occurrence of one event in space influences the positioning of other events in space. A good example here would be the spread of infectious disease. In 17th century Huronia, for example, once small pox had been introduced to a village, there was a high probability that many other cases would occur there as well. In examining the distribution of various artifact types on an archaeological site, we might notice that not all types are randomly distributed over the site. Projectile points, preforms and flakes of bifacial reduction might tend to occur together, whereas scrapers may be located separately and pottery might be located differently from the others. The tendency for certain types to cluster with each other and potentially with other artifact types is a second order effect.
Archaeological analysis along these lines informed the basis of much of the application of spatial statistics towards determining activity areas in the 1980s, despite the fact that the first/second order effects terminology was not used at that time.
In archaeology the majority of the material with which we deal is clustered. A site is most frequently a cluster of artifacts occurring somewhere on the landscape, surrounded by adjacent areas with no artifacts or, at least, significantly reduced numbers of artifacts. Thus, at one level of analysis, the cluster of artifacts representing a site can be considered a second order effect since, once the site location is selected, the occurrence of artifact locations will always be near other artifact locations. At a different level of analysis, when it comes to analyzing the relative distribution of various artifact types on a site, it is better to treat the overall distribution of all artifacts as the first order effect and the
relative positioning of various types within that as a second order effect. However, while the site selection itself on the landscape based on specific preferred topographical
features would be a first order effect, site selection based on close spatial proximity to other closely related human groups would be a second order effect.
So far we have discussed the interaction of second order effects as an attractive process leading to clustering of events. However, O’Sullivan and Unwin (2003:65) use the example of 19th century supply towns across the Canadian Prairies as an illustration of a second order effect, where one event precludes the presence of others nearby. Here the positioning of one town effectively suppressed occurrences of other towns in close proximity, leading to an overall pattern where the towns tend to be evenly spaced over the landscape with the distance between them related to the economics of travel time to get to a town. Of course, such an effect is at the centre of many classic geographic models, such as those that employ Central Place Theory (e.g. Christaller 1972).
From the preceding discussion, the boundary between first and second order effects can be somewhat fluid, especially in archaeology. For example, site selection might be a first or second order effect or possibly both. In the case studies that are included in this dissertation, a collection of material from the surface of an Ontario Late Archaic site and an ancient Egyptian cemetery, the analyses treat the actual distribution of events over the
site as a first order effect but focuses on the relationship between the events, which is truly a second order effect. In any event, the distinction is a useful one to make, even if somewhat arbitrary.
Another aspect of the analysis of second order effects is that they occur over a distance less than the size of the study area. There is an upper limit to the distances to be
considered and this distance should be small in relation to the overall size of the study area. While this was observed in practice, it also seems logical since we are looking for second order effects that should occur at distances well under the overall size of the study area. Larger distances thus become meaningless. The specific distance most likely varies with the nature of the second order effects being examined. With the Kellis-2 cemetery, it was found that 3, 5 and 7 m were practical sizes, 10 m was problematic and 15 m and over seemed to be meaningless. In order to quantify this result, a value of 10 m is just over 20% of the square root of the site area. Whether this result would hold in other cases is unclear.
For further discussion of first and second order effects the reader is referred to O’Sullivan and Unwin (2003). There is also a description in Bailey and Gatrell (1995). However, it is somewhat confusing and I question whether the example used properly describes the various effects and how they differ.