Kettle Falls

I am heading up to Kettle Falls, Washington today. Was invited to a ceremony by the Colville Tribe. As I worked along the Upper Columbia every spring from about 1997 to 2003, it will be like going home again. It was definitely one of my great jobs in archaeology, although I wish I knew then what I know now. Working up there was like archaeology camp, archaeology all day, every day, with nothing else to do but drink (at least for the crews). The area has been built up these days, but back then there was no or limited internet and cell phone connections were very sporadic. Houses were scattered and not many people were around. There was no one to bother you, no way to really get a hold of you. It was wonderful. I think crews today don’t know what they are missing by having technology.



Lake Roosevelt, the reservoir behind Grand Coulee Dam, is huge. Its 150 miles long, averages about one mile wide. At the dam the lake is over 400 feet deep, while at Kettle Falls its still about 80 feet. For those who don’t know, the entire reservoir zone was stripped of vegetation and buildings prior to filling. When the water is low, entire denuded landforms are exposed. It’s pretty amazing to see all the ridges and swales with no trees or brush. Reservoir sediments have in filled many low areas, but the landscape’s basic form is remarkably visible.

There are sites everywhere. Prehistoric and historic. I, of course, was most interested in the prehistoric stuff. That’s what my thesis was focused on. Lake Roosevelt land use. Not sure I learned much about LR land use, but I certainly feel like I learned a huge amount about archaeology, what to do, what works and what does not work.

I have to revisit my thesis data. My results were somewhat disappointing, mainly because I was not quite sure why I could not identify functionally differentiated sites. Are the sites just not that different? That is possible. Did I do something wrong? That is also possible. Is the data just in bad shape and not usable? Again, that is also a possibility.

I suspect, however, that a combination of all these factors influenced the results. One of the biggest problems was that everything seems driven by assemblage size (surprise, surprise) so my inability to normalize the data was a problem. Volume and area excavated was rarely provided and overall site size is sketchy as most sites are so poorly recorded.

I think I need to do several things with the data. First, instead of using all data collected, I need to select out those cases where the data is in good shape (volume, areas excavated, good site size, etc.) and use only these in the analysis. This many mean that about 70% of all data is not used, but not sure I have a choice. I should also drop those sites that date to broad time periods (i.e. 8,000 to 150 bp) and only use those assemblages that have tight time frames and date to a single temporal period. This would reduce the problems caused by palimpsests.