More on WARP Data (Part Two)

Yesterday, I wrote about “Densities and Visibility” and “Hidden Landscapes” with regard to the data generated by the Western Argolid Regional Project. Today, I am going to write up four more aspects of our virtual study season in what should be the final installment of WARP related writing this summer. (You can read the rest here, herehere, and here). 

Next week, I return to Cyprus (at least in my writing and reading), but for now, WARP is the place.

Here are the final four observations on the WARP data crunching season.  

3. Land Use. As part of our standard descriptions of each survey unit, we recorded a good bit of land use data. This includes things like dominant vegetation, evidence for recent plowing, and the presence of features such as terrace walls that indicate material investment in the landscape. We initially chose to record this kind of information to provide insights to artifact recovery rates, but we soon discovered that this data also provided a high resolution perspective on contemporary (and recent) land use in the Inachos valley. 

For example, it became clear from our data that three main commercial crops in the Inachos valley were olives, stone fruit (primarily apricots), and citrus. While olives are more or less ubiquitous throughout the survey area, citrus tend to only appear in units under 150 masl in elevation. Apricots appear in units under 200 masl leaving units of 250 masl and higher in elevation to olives. This likely has to do with the susceptibility of these crops to frost damage. The presence of windmill-like air circulators in the citrus fields to the northwest of Argos confirm that this territory receives some manageable frost during the winter months. More durable apricots, a major export crop in our survey area, can endure occasional frosts, but are less rugged than the ubiquitous Greek olive tree. 

Evidence for plowing tends to be most common at fields under 200 masl which have, for our survey area, moderate slopes of less than 13 degrees although fields with compacted soils that show some evidence for plowing in the recent past (which we call “plowed, compacted” soils) extend slightly higher in elevation (and average 215 masl) and with slightly greater slopes of an average of 14 degrees. Higher elevations and greater slopes than these tend not to see regular plowing and are characterized by unplowed field even when the erosion of soils on slopes exceeding 20 degrees creates loose soil conditions. It should come as little surprise that units with plowed and loose soils tend to have the highest artifact densities and the highest visibilities. It is worth noting, however, that units with plowed and compacted soils, which indicate recent, but not ongoing plowing, produced the higher densities than predicted by visibility alone. 

A significant network of terrace walls manage the sloping walls of the Inachos river valley, and we recorded over 850 units with terrace walls. This is over 10% of all survey units. The average elevation of a terraced field is 247 masl with no terraces appearing below 108 masl and the highest over 500 masl. The average slope of a terraced field was 26 degrees. Predictably, most of the higher terraced fields (over 245 masl) were not plowed and these tended to have steeper slopes. There were, however, a number of recently or currently plowed terraced fields at lower elevations and slightly lower slopes suggesting that access rather than elevation and slope alone determined whether terraced fields saw plowing. In fact, these recently or currently plowed terraced fields tended to produce much higher artifact densities than visibility alone would predict whereas unplowed terraced fields tended to perform closer to what one would expect based on their visibility. This almost certainly reflects the high artifact densities from fields surrounding the ancient acropolis of Orneai which is also in the immediate vicinity of the village of Lyrkeia and accessible via a network of paved and field roads.      

4. Describing Chronological Landscapes. Over the last week or so, the project directors have been thinking about how best to describe the distribution of material from various periods across the entire landscape. This is distinct from how we interpret or understand the historical significance of particular patterns in the landscape. Instead, the idea (to my mind) is to describe the distribution of material in a consistent way across the entire survey area that allows for at least basic comparisons.

On the most basic level we can compare the character of assemblages by number of artifacts alone, but this speaks very little to the distribution of artifacts across our survey area. Thus combining the number of artifacts with the area of the units in which they appear helps to give some sense of distribution. David Pettegrew in his recent (unpublished) analysis of the distribution of EKAS data used nearest neighbor analysis (based, I believed on the centroid of units) determine whether the pattern produced by artifacts from various periods is clustered or dispersed. The vagaries of artifact recovery patterns could, I imagine, be managed by comparison with the overall pattern of the survey which would allow us to say whether the overall distribution of artifacts from a particular period is more or less dispersed than the overall distribution of all artifacts from the survey (imagining that the latter reflects recovery rates). 

Obviously one challenge here is the differential visibility or diagnosticity of particular periods on the surface. Certain periods – such as Pettegrew famously argued for the Late Roman period in Greece – are more visible than others complicating a simple reading of distributional analysis as a measure for (say) the character of settlement in the survey area. The other challenge, of course, is the different date ranges for various periods which mean that comparing, say, the Late Roman period (which we date to AD300-AD700) tends to be a good bit longer than, say, the Classical period (450BC-300BC) which means that the Late Roman assemblage has had twice as long to develop in the landscape.

There are various ways to manage for the differential diagnosticity and the different length of various periods to make these assemblages comparable. I tend to be fairly pessimistic about the potential of comparing assemblages from different periods. In other words, I think it is pretty hard to make arguments for the expansion or contraction of settlement by comparing assemblages from two different periods unless one establishes that the material signature of the two periods is fundamentally comparable.

That said, I suspect that the distribution patterns of material from various period between different survey projects is likely to be more comparable than between periods in the same survey project. For example, issues of differential visibility or diagnosticity on the surface tend to be common to most survey projects in a region and in most cases periodization schemes are, if not absolutely the same, at least broadly consistent at a regional level. In other words, being able to describe the various period landscapes across the survey area serves as the basis for later analysis of the periods in question rather than the analysis, necessarily, of the survey area across time (although it should also inform how we understand the survey area diachronically).  

5. Chasing the Data. One thing that crunching data does reveal is the strengths and weakness of any dataset. Our dataset is quite a way from what I would consider big data and as a consequence little problems with our data can create big issues during analysis. (And here I’m assuming that the strength of big data schemes is that small imperfections or outliers in the data set tend be washed out by the scale of the data more generally, for better or worse). As I ran queries and did analyses and produced new datasets on the basis of data that we collected in the field, I discovered little problems. For example, the aoristic analysis that I posted last week was based on a chronology table that had the Archaic period dating from 750BC-AD450 rather than 750BC-450BC. This is meant that pottery dated to the Archaic period was rather significantly underrepresented in the aoristic analysis that I conducted. It is an easy enough fix, fortunately, one that probably would have become clear at some point in the publication process.

At the same time, doing the work of analyzing our material is part of what brings various limitations to our data to the fore. For example, we didn’t ask our field teams to record the presence of terrace walls. So I had to excavate this data from the a more general comment field. This was easy enough to do, of course, but I suspect that the dataset is a bit fuzzier around the edges than one generated by a simple check box. 

In the end, querying the data will both reveal its analytical limits and make it a stronger dataset. This kind of “slow data” work is both humbling, in that it reveals the limits of data collection processes in the field, and energizing in that it only through analysis do we recognize the potential of our data to reveal more about the landscape than we had intended.

6. Solitary Data Crunching. Finally, crunching data by myself has been pretty boring. One of the great things about study seasons is not so much the work of study, but the time to reflect, ask questions, make false starts, share processes, and think out loud (although my colleagues might not entirely agree about that last one!).

Crunching data alone in my home office feels so disconnected from the work of the survey. I’m left to my own devices and my own questions, I often end up spinning my wheels or working my way into a dead end of data which does neither speaks to whatever hypothesis that I have imagined nor leads me to new questions. 

When doing data crunching next to my (often much smarter) colleagues, however, I constantly encounter new ways of seeing the data and imagining how it speaks to the archaeological landscapes that we explored together. In that context, data oriented study seasons often led to trips through the survey area (and surrounding regions), shared memories and reflections on units and field practices, and deeper engagements with both the landscape and our data.

Data-ing alone, on the other hand, has made me feel not only a bit detached from the survey universe, but also mildly confounded by our data. Hopefully before we get to the publication stage, we’ll have time to revisit our data together in a more collaborative and conversational way, but for now, this is what we have and despite it being a bit uncomfortable, I think I’ve made a bit of progress. 

More on WARP Data (Part One)

As readers of this blog know, my colleagues and I on the Western Argolid Regional Project have been working our way through the data that the project produced over its three main field seasons. This (virtual) study season focused on producing some reports that would start to describe the survey area in general ways. These reports will help us in the future as we attempt to unpack the character of more specific artifact scatters and distribution patterns. They’ll also contribute to address some “big picture” questions in survey methods and methodologies. I’ve already blogged a bit on my work over the last few weeks (here, here, and here). This week’s installment will likely be my last and the most general.

Six things that I learned this WARP study season.

1. Densities and Visibility. Like most intensive surveys we counted the total number of sherds and tile fragments in each survey unit. These numbers alone don’t tell us much about the past, but they do provide insights into artifact recovery rates across the entire survey area. For example, we know that surface visibility and the quantity of artifacts in the plow zone are likely independent variables. The question then becomes whether the areas with higher artifact densities represent the quantity of material in the plow zone or simply zones where we have higher visibility (or geomorphological processes took place that revealed past surfaces). As I blogged about a few weeks ago, for our survey area lower visibility tended to mean lower densities in a fairly consistent way. This indicates that across the entire survey zone lower visibility areas are as likely to have significant quantities of artifacts as higher visibility areas (and vice versa). In other words, it suggests that the two variables are independent. We might see a different trend if the two variables were related. For example, if ancient settlements tended to appear more frequently in fields that were currently abandoned and overgrown, we might see unusual spikes in artifact densities at lower surface visibilities and lower densities – either proportionately or in absolute terms – with greater visibility.  

Another measure for this considers the variability of artifact densities at different levels of visibilities. A number of scholars have noted that as visibilities decrease our ability to consistently recover artifacts becomes more random. This may be the case in some environments, but at WARP, the variability present across visibilities (measured by the standard deviation divided by the mean) does not pattern consistently and if anything appears to slightly increase as visibility increases. 

VarVis

Background disturbance is another factor that archaeologists have recognized as impacting artifact recovery rates (as I’ve blogged about before). Background disturbance describes the amount of distracting material on the surface that complicates artifact recovery. In general, we did not notice background disturbance having a significant impact on artifact recovery rates at WARP in general. That said, when consider variability in artifact densities per visibility split out by background disturbance (which we recorded as light, moderate, and heavy) we can see that variability decreases with higher densities, but the pattern is not a particular strong one. It is telling, however, that variability increases with visibility in units with moderate and light artifact densities.

VarBGD

While the patterns here are not particularly compelling, I think they do hint at how background disturbances shape recovery rates even if the story they tell is not a dramatic one.

2. Hidden Landscapes. Survey archaeologist have long worried about hidden landscapes, a phrase coined by John Bintliff, Phil Howard and Anthony Snodgrass in an influential late 1999s article. While this article introduces a whole range of reasons for landscapes being hidden – from geomorphology to the fragility of ceramics fired at lower temperatures at certain periods – in crunching the WARP data, I concerned myself with just one or two factors. The main factor that I fretted about was the possibility that contemporary land use which shapes not only recovery rates (through factors like surface visibility and plowing; units that remain under cultivation, for example, tend to have better surface visibility and to receive the kind of attention that increases the amount of subsurface contexts visible to field walkers), but as importantly shaped access to fields. Overgrown fields, steep slopes or units inaccessible via field road, walking paths, or undergraduate clambering may well hide landscapes that survey projects could systematically overlook. At this point in our project, there is no chance of us going back to field to survey steep slopes or dense thickets of maquis, but we can look at the data to see if there are hints that certain periods were more common in certain kinds of fields or, better still, whether they appeared somehow out of sync with predominant densities across the survey area. Again, overall artifact densities do not necessarily suggest certain units saw more activity throughout the past than others, but suggest the happy confluence of field conditions and surface material.

To do this analysis, I ranked the units in our survey area by artifact density according to equal intervals. The lowest density unit was 1 and the highest was 5. I think looked at any period of under 1100 years (and these constitute “narrow” periods from WARP) and ordered them according to average density rank. The idea was to determine whether certain periods appeared consistently in fields where artifact recovery rates produced low densities. I recognize that this is not the only method to suss out whether certain periods were more hidden than others, but it offered at least a preliminary way to consider the problem at scale across our survey area.

For Bintliff and co. there was a thought that certain prehistoric periods might be hidden in their landscapes in Boeotia. On WARP, however, it appears that prehistoric periods – even rather narrow ones such as Late Helladic IIIA-B – tend to appear most consistently in units with the highest artifact densities suggesting the confluence of recovery rates and material in the plow zone revealed a significant assemblage of pre-historic material. Of course, this can’t tell us what we didn’t find, but it is suggestive when compared to material from the Medieval period, the Venetian/Ottoman period, and the Early Modern period. This material tended to appear in units with lower average densities and lower average visibilities suggesting that contemporary land use, access, and surface visibility might be working to obscure the Late Medieval and post-Medieval landscape in the survey area. The densely overgrown Medieval site near the village of Sterna which produced Late Medieval through Venetian material despite may be a useful example of an area where our survey might not consistently explore. The site stood above the highest cultivated fields in the Inachos valley and was densely overgrown in vines, trees, and weeds. Artifact densities was spotty owing to the difficult surface visibility and it seems probable that artifact densities if not otherwise compromised would have been significant. That said, the site near Sterna did produce some standing architecture, a series of (you guessed it!) cisterns, and stood at a strategic location where the Inachos river valley narrowed before opening out into the wider Lyrkeia valley. In other words, there are certain characteristics of the site that encouraged us to explore it in greater detail despite the limited visibility and challenging landscape. To my mind, this mitigates to some extent against the idea that the absence of certain landforms and situations have led to the underrepresentation of Late Medieval to Early Modern material, but this hypothesis will, of course, require greater testing.

Stay tuned for more tomorrow!

Comparing Assemblage Part 2

Yesterday, I introduced some of the work I’ve been doing with data from the Western Argolid Regional Project. I compared artifact assemblages from the same unit, but produced by two different methods: our standard survey field walking and more intensive 2 m radius total collection circles. 

As we have noted from similar experiments with the Pyla-Koutsopetria Archaeological Project, artifact densities produced by total collection circles tend to be significantly higher than those produced by standard survey. What remained less clear, however, is whether the larger number of artifacts recovered per square meter add significantly to what we know about the material on the surface or whether they mainly trade in redundant data.

More specifically, I was curious whether more intensive sampling of units with lower surface visibility might help us open a larger window into the assemblages present in the plow zone. This may seem more or less intuitive except that historically survey projects have tended to increase intensity in areas where more artifacts appear. Typically, projects employed more intensive collection strategies which ranged from gridded collection to total collection circles for areas designated as “sites.” 

To think about this, I compared pairs of units from the same area with different visibilities. For example, units 3420 and 3421 are from a large Roman site in the Lyrkeia Valley. Both units had above average artifact densities with 3420 having rather exceptional sherd densities (over 6000 sherds per hectare!) and the unit 3421 having high artifact densities (920 sherds per hectare). Unit 3421 has a visibility of 50% and unit 3420 was 80%. The areas selected for resurvey in both units had the visibility of 90%. They both had light background disturbance and compacted soils. The biggest difference between the two units is their size. 3420 is 2200 square meters and 3421 is smaller at 860 square meters. 

UntitledImage 

The standard survey of 3420 produced 285 artifacts and the Resurvey 1 and 2 produced 83 and 115 respectively. The big spike in the standard survey line represents a gaggle of Early Roman amphora sherds turned up by the standard field walking. The slight hump in the blue line between 2600 BC and 2300 BC is from an EHII pithos sherd. Otherwise the profiles produced by the two types of survey are remarkable similar with the standard survey producing a bit more material datable to narrow periods.

UntitledImage

The results are largely the same from unit 3421. The most obvious difference, of course, is that Resurvey 2 produced 131 artifacts which is more than the 43 produced from standard survey and 79 from Resurvey 1. Despite the difference in quantity, the profiles is remarkably similar. Resurvey 1 recovered a small group of sherds of Bronze Age – Iron Age date leading to the orange line starting around 3200BC. The combination of material of Classical-Roman, Roman, and Late Roman dates recovered in Resurvey 2 produces the double-humped line that dominates this profile. The parallel, if lower lines, produced by the Resurvey 1 and Standard assemblages demonstrate that the increase in quantity adds little to the chronological range or distribution of material found in this unit.

Another case study produces somewhat different results. Comparing two units, 11027 and 11033 from the same area with lower visibilities, 40% and 30% respectively and slightly lower densities (which are nevertheless quite high for our survey area) of 207 sherds per hectare and 356 sherds per hectare respectively. 11027 has moderate background disturbance and 11033 has light and both have low vegetation and compacted soils. The biggest difference is unit size with 11027 being a positively massive unit for our survey at over 4800 square meters and 11033 being rather average with a size of 1684 square meters.

UntitledImage

The profiles produced by the various survey methods employed in unit 11027 are quite different. Standard survey produced an assemblage of 35 artifacts, while Resurvey 1 collected only 20. The assemblage from Resurvey 2, however, was 57 artifacts. The increase in material starting at around 1400 BC is driven by a single LHIII kitchen ware sherd and sustained by a robust, but rather undiagnostic collection of “Ancient Historic” (material dated to between 1050 BC and AD 700) semi-fine and kitchen wares that is paralleled at lower intensity by “Ancient Historic” material from Resurvey 2.  Both Resurvey 2 and Standard survey produced a spike in Late Roman and Roman utility and kitchen wares. Significantly, however, Resurvey 1 collected 3 sherds datable to the Archaic-Hellenistic period and material from these periods did not appear in the standard survey.

UntitledImage

Unit 11033 produced generally smaller assemblages with 23 sherds from Standard survey and 8 and 22 from Resurvey 1 and Resurvey 2 respectively. The major spike from Resurvey 1comes as a result of four semi-fine wares datable to the relatively narrow Early Roman period and is paralleled by a lower spike from the Standard survey. Resurvey 2 produced no other sherds datable to any period shorter than 1600 years! Resurvey 1 and the Standard survey produced some Classical-Hellenistic, Hellenistic-Roman, Roman, Late Roman, and as the late spike in blue shows, Medieval, Ottoman-Venetian material. The hump visible in the Resurvey 2 profile dating to around 450 BC represents a single Classical-Hellenistic fineware sherd and a collection of 13 semi-fine ware sherds dated to the rather broad Classical-Roman period. Despite the rather different profiles, then, the major patterns produced by diagnostic material are surprisingly similar between the two units.  

It is a bit tricky to generalize from two case studies, but the second pairing of units suggests that more intensive collection strategies when applied to units with low visibility but high artifact densities can produce additional chronological information from the units. In the first example in which visibility is higher, the Standard survey and resurvey profiles tend to be more similar.

Again, this is a very small case study, but it is suggestive that the traditional practice of increasing intensity in areas with high artifact densities (sometimes called “sites,” or euphemistically, “LOCAs” or “POSIs” or whatever) is less likely to produce new chronological information than a similar approach to units with lower surface visibility and correspondingly lower artifact densities. We did quite a bit of resurvey work in quite a few contexts across the WARP survey universe and by comparing the assemblages produced by different methods using Aoristic analysis, we should be able to test this hypothesis a good bit. 

It goes without saying that chronology is just one indicator of significance for artifacts collected over the course of survey. Variation in fabric, in function, and in shape also informs the analysis of surface assemblages and Aoristic analysis does not account for variation in these areas. At the same time, chronological continuity is the sine qua non for most analysis of survey material. If you can’t date material, then it is harder to make any historical arguments for it.

It is worth stating that this analysis is very preliminary and I am going to continue to tweak and firm up my formulas and the underlying data. And, we do plan on making all the data that supports this analysis available openly via Open Context

Comparing Assemblages

Over the last few weeks, I’ve been working with the data collected from the Western Argolid Research Project. Mostly this has involved going through the survey unit data that we collected as a way to determine general patterns of artifact recovery rates from across the survey area. This is important work, but largely unrewarding. I had run many of the queries on the fly while survey teams were still working in the field and had started to recognize the general patterns as they were emerging in the data. That said, it was still necessary to check things with the complete data set and I’ll post some more on this next week.

For today, I want to post on another aspect of my work with the WARP data. Over the three main field seasons, we conducted a series of revisits to our survey units to collect additional samples of material from the surface. Teams did this by collecting all artifacts in a circle with a 2 m radius from the survey units. We selected units for resurvey on the basis of location within the survey area, surface visibility, and artifact densities. 

These revisits had two main goals. First and foremost, we sought to use them to calibrate the assemblages produced through our standard survey methods where we collected density data and artifacts from a 20% sample of the surface through traditional field walking techniques. Secondly, we hoped that these total collection circles would also provide us with more robust samples from units where visibility, for example, compromised artifact collection.

I had started messing with this data in 2018 and made only a tiny bit of impressionistic progress. The biggest challenge for me was figuring out how to compare the three assemblages in a useful way. I discuss some of my difficulties here.

The biggest challenge was dealing with various chronologies that we assigned to our ceramics. The chronotype system allowed our ceramicists to assign a range of chronologies (and dates) to artifacts recovered in the survey. In some cases, these were immensely broad (e.g. “Ancient Historic” which has dates from 1050 BC to AD 700) and in some cases they were particularly narrow (e.g. Roman, Early which dates from 50 BC to 150 AD) and, in many cases, they were of more middling resolution (e.g. Roman which dates from 50 BC to AD 700). The different assemblages produced artifacts not only from different periods, but also from periods with different chronological resolutions. So answering the question whether more intensive collection strategies, such as total collection circles, produced different or just more material had to take into account the different resolutions at which we identified artifacts. In other words, we had to figure out whether recovering material dated to the “Early Roman” period in a resurvey circle mattered if we collected material datable to the broader, but inclusive “Roman” period in standard survey.

The solution to this problem was doing a little Aoristic analysis as a standardized way to represent the chronological profile of different assemblages. Aoristic analysis assumes that all artifacts have an equal chance of appearing in any year of a given chronological range. In other words, an artifact dated to the Roman period has an equal chance of appearing in any year between 50 BC and AD 700. Artifacts from more narrow period have greater values per year reflecting the increased likelihood that they would appear in any given year in the span. To complicate matters a bit, I also factored in the number of artifacts from any particular period. As a result, the charts that follow reflect not only the likelihood that an artifact dates to a particular year, but so the quantity of artifacts present in the assemblage with any probability of appearing in any particular year. I recognize that this is combining apples (that is probability of any artifact appearing in a year) and oranges (the quantity of artifacts present with various probabilities), but if we are using Aoristic analysis and its corresponding visualizations as a heuristic, then this kind of conflation is maybe a reasonable way to combine various kinds of data into one chart.  

Let’s look at some data. For the unit 2517, which is near the acropolis of Orneai and had 30% visibility, we did standard survey and two total collection circles. The standard survey produced 89 artifacts and the resurvey units produced 77 and 146 respectively. Despite the differences in quantity, the assemblages had rather similar profiles with the only exceptions being a bump in the Early Bronze Age generated by a small number of rather diagnostic EHI-II and EHI sherds. Resurvey 1 produced a Late Roman and an Early Medieval sherd which created the slight bump in the orange line. The presence of artifacts dated to the Archaic-Hellenistic period in Resurvey 2 created a more nuance profile in the centuries prior to the notable Classical-Hellenistic spike. The general similarities of the two profiles reflects the basic similarity between the three assemblages, but it is clear the material from resurvey did provide chronological nuance to the 

UntitledImage

Another unit from the same area, 4345, had three resurvey circles. Unlike unit 2517, it had higher visibility and a correspondingly more robust assemblage with 434 artifacts recovered during standard survey and 45, 173, and 68 recovered from three resurvey circles.

Similar to unit 2517, the basic profile of the assemblages appear the same, but the resurvey units do produce some notable spikes. Resurvey 2, for example, produced a single Late Helladic IIIA sherd. Resurvey 3, recovered 6 artifacts dating to the Archaic-Classical period and one Archaic sherd producing a narrower spike than the material dating to the Archaic-Hellenistic period recovered in the standard survey and in Resurvey 2. Resurvey 1 produce a single piece of Ottoman semi-fine ware which has a narrow 120 year date range and produced a late spike in the chart.   

UntitledImage

I have more case studies withs smaller assemblages that I’ll share later this week, but for now, this kind of visualization seems to be a useful way to compare groups of artifacts produced by different methods from the same units. What this will reveal about survey methods, both in general and in different circumstances, remains the question that more analysis will hopefully answer!

Preliminary Thoughts on Artifact Recovery Rates from the Western Argolid Regional Project

This past week, I’ve started the intimidating task of crunching the data produced over three field seasons with the Western Argolid Regional Project. While we’ve made a few efforts to make sense of the data over the past five years, our dataset has been varying degrees of provisional and more pressing matters in the field and in the storerooms often attracted our attention. With the field and storeroom over five thousand mile away and our data as clean as any project can reasonably expect, now is the time for number crunching! 

In the past, we have tried to focus on a number of rather well defined publication projects: a preliminary report and various side projects that required some attention. This year, we wanted to shift our attention back to analysis and instead of producing fully formed publishable quality manuscripts, we wanted to produce some reports and moved the project forward without the pressure of polished final publications.

This summer, I elected to look at the variables that shaped artifact recovery in the field with the hope that this might inform how we analyze artifact patterns in the landscape. So far, I’ve just started but I can make a few observations (and these, if I recall correctly, largely follow observations that I made several years ago when analyzing a rougher version of the same data).

First, the most significant variable in artifact recover is surface visibility. Survey archaeologists have know this for years so it comes as no surprise. It appears that sherd density tracks pretty closely with density up to the highest visibility units (100%) where densities drop rather steeply (as does sample size!).

WARP Charts  Google Docs 2021 06 08 09 43 32

WARP Charts  Google Docs 2021 06 08 09 44 16

 

 

Tile densities track visibility a bit less regularly and follow a kind dromedary curve with a hump at 40% visibility and another peak at 90%. The reason for this is a bit unclear. It may be that tiles are generally a bit more visible in the plow zone so surface visibility doesn’t impact their recovery quite as dramatically. A good example of this is that many of the highest density units with tile are from the immediate vicinity of collapsing houses at Chelmis and Iliopouleika (6 of the top 10 and 13 of the top 20), and these units tend to have visibility below 50%. In these units, tiles are abundant and often fairly well preserved and this likely contributed to their relatively high recovery rates even from units with lower visibility.

Second, my old buddy David Pettegrew has been running similar analyses on the EKAS data (which is rapidly becoming available at Open Context). Of particular interest to him (and to us!) is the impact of background disturbance on artifact recovery rates. As we say in the WARP field manual: this category represents the degree to which a field walker’s ability to see artifacts on the ground is hindered or obscured. This is a distinct category from visibility since even a field with 100% visibility could still have heavy background disturbance. A useful rule of thumb is that when walkers are spending much of their time picking up rocks they think are pieces of pottery, the background disturbance is heavy.

There are any number of ways to measure background disturbance. For example, units with high background disturbance took about 2 minutes longer to walk than units with moderate or light background disturbance despite having an average visibility of 68.5% as compared to 47.3% and 57.0% for light and moderate background disturbance respectively. Units with high or moderate background disturbance had a tendency to produce more “Stone, Unworked” (which are really just rocks) than those with light or none (2.8 and 2.3 rocks from units with high and moderate background disturbance and 1.9 rocks from those with light and none). 

On EKAS, there was a relationship between background disturbance and artifact recovery rates. In fact, David has proposed a metric that takes into account background disturbance and visibility to understand recovery rates in those units (and he has plans to unpack some of this in a future publication). That said, when we analyze the background disturbance from the Western Argolid, it doesn’t seem to have a particularly strong relationship with recovery rates at least as manifest in artifact densities. 

For units with the heaviest background disturbance (n=672), in fact, artifact densities tracked more or less along with those from similar visibility units with the exception of two spikes at 40% (n=33) and 90% visibility (n=88) where units with heavy background disturbance produced higher densities than might be expected from visibility alone. In contrast, units with moderate and light background disturbances more or less followed the expected trajectory based on visibility alone. This suggests that background disturbance did not exert a predictable influence over artifact recovery.

WARP Charts  Google Docs 2021 06 08 09 45 08

WARP Charts  Google Docs 2021 06 08 09 45 59

WARP Charts  Google Docs 2021 06 08 09 46 42

We obviously recorded more variable than background disturbance and I have began to run quarries on our data that looks at these variable as well. So, if you’re a survey archaeology “method-head” you might want to stay tuned for more “exciting” methodological reflections in the coming week.

In the meantime, I also ran some queries based on artifact recovery and vegetation in our units. We had standardized recording terms for vegetation in each unit which ranged from “weeds,” “maquis,” and “phrygana,” to “citrus,” “olives,” “grain,” and “grain stubble.” It was possible to select multiple vegetation types for each field resulting in 27 combinations which appeared in at least 50 units. Various combinations produce artifact densities that under performed what one might expect from visibility alone.

The lowest visibility were typically flat units lower elevations (< 200 masl) with citrus or stone fruits (and not infrequently weeds). My guess is that these units were as likely to be shaped by their proximity to the Inachos River and its wandering course that deposit sediments carried toward the Argolidic Gulf. In contrast, units with higher slopes and elevations, often populated with olives, weeds, and (mostly volunteer) grains produced artifact densities that exceeded those predicted by visibility alone. This is as likely the result of historical phenomena as artifact recovery variables and shaped by the dense scatters associated with the fields around the acropolis of Orneai.

As you might guess, such hypotheses will have to be tested using our GIS data, but for now, I’m mostly just crunching numbers without too much attention to spatial concerns. Once again, this means more “method-head” goodness is likely to appear in these pages in the near future!   

WARP 2021 Study Season

The 2021 WARP study season starts tomorrow. This means three things.

First, it means DATA. Like many contemporary archaeological projects and certainly most contemporary surveys, WARP produced a ton of data from its four seasons in the field and three study seasons. Despite spending some quality time with this data each year, it remained a bit provisional as our finds data was refined and updated and our survey unit data was polished. Moreover, as we digitize and analyze maps, we continue to produce more data that can inform our larger analysis. In short, this means a season of sitting in front of my laptop and crunching numbers.

Our biggest goals this season is to determine the main factors that impact artifact recovery rates from our survey area and then attempt to determine whether the variables impact recovery rates in the same way for artifacts from every period. 

Second, it means DISPLACEMENT. Some of my fondest memories of archaeological work do not involve toiling in a trench or slogging through another field looking for sherds. They don’t even involving hiking up a mountain and the rush at “discovering” an undocumented or unpublished fortification. Some of my favorite memories of doing archaeological work involve sitting at my laptop in the tiny room underneath the Marinos house in Ancient Corinth, crunching EKAS data with David Pettegrew. I also have fond memories of working on Polis data on Cyprus while sitting in the Polis storerooms or in the main room of our little apartment in the village.

In both of these cases, we had the ability to go out the door and wander around the excavation area or go and check out a particular unit, situation, or view. I’ve never been one for aimless driving around or hiking or other random outdoorsy activity that I don’t perceive as having a clear goal in mind. I do enjoy, however, checking things out and revisiting sites or scrutinizing problems at a site or in the landscape. The dialogue between the data and sites and landscapes ensures that the data remain tied to experience. In fact, I often think of data that we take with us into the field (either in our minds or quite literally when we check a measurement or test a hypothesis) as embodied data. These data are data that blend seamlessly with the sites themselves.

Of course, this year, like last, we can’t do that. I’m feeling a distinct sense of displacement from the field and it reinforces my idea that data as data, set adrift from a sense of place, loses something significant. 

Finally, no study season can happen without DONUTS. Tomorrow is National Donut Day. My plan is to make a donut pilgrimage to Sandy’s Donuts in Fargo to mark the official start of the WARP study season. 

WARP Field Manual: A Manual for an Intensive Pedestrian Survey

Over the last month or so, I’ve been puttering around with the field manual from the Western Argolid Regional Project. This was an intensive pedestrian survey conducted in the Inachos River valley from 2014-2016 (with study seasons in from 2017-2019).

We produced a field manual that we then updated as the project went along. In an effort both to contribute to the small number of publicly available field manuals from field projects and to make our project a bit more transparent, we decided to tidy up our manual and make it available via tDAR.

Some of my long-time readers might remember that a few years ago, I was keen to formally publish as many field manuals as I could via The Digital Press at the University of North Dakota. We formally published on field manual, the iconic Corinth Excavations Archaeological Manual in 2017 and it has been a solid and consistent performer, download over 1000 times and used in any number of university and college classrooms. We also prepared a little archive of archaeological field manual and you can explore it a bit here.

This work generated some tepid interest in formally publishing field manuals, but nothing came of it. In fact, even my WARP colleagues were pretty ambivalent about publishing our manual. I did typeset the WARP manual together so that if someone wanted to publish it, they could. We also made it available under an open access CC-By license.

In any event, you can download the WARP manual here. It’ll be up in tDAR in the next week or so and I’ll share that link as well.

WARP COVER FINAL01

Finalizing a Survey Field Manual

A few years ago, I casually floated the idea that projects should publish their field manuals. This was in conjunction with the publication of the Corinth Excavations Archaeological Manual (by Guy Sanders, Sarah James and Alicia Carter Johnson) by The Digital Press at the University of North Dakota. There was a pretty tepid response with a number of project directors agreeing that this was a good idea in theory, but no one took me up on the suggestion and submitted a manuscript.

I’m still very open to the idea and I’d love to publish a manual from any of the iconic excavations in the Mediterranean! Field manuals represent the crucial link between methods (and methodology) and field practices that often have a significant impact on the kind of knowledge a project produces. They also provide insight into project and situation specific constraints, offer a kind of paradata (as well as metadata) for the project’s data, and give some indication of the work conditions and work rhythms present on site. Manuals also have pedagogical value as both evidence for how students learn archaeology on the ground and as examples in the classroom for how methodology plays out in the field. Finally, a publicly available field manual provides the kind of transparency that is good practice for the discipline. 

As part of The Digital Press’s project to publish the Corinth Excavations Archaeological Manual we also published an archived list of project manuals which is available here.

Part of the challenge, of course, in publishing a field manual is that field manuals tend to be dynamic documents that change over time. Even for a relatively short project, such as our Western Argolid Regional Project, the manual underwent a number of changes over its four seasons of use. We were particularly fortunate to have active and engaged survey team leaders who provided not only input into the manual itself, but also helped us revise it each year. As a result, publishing a final manual is not as simple as just formatting a document and sending it to an archival repository like tDAR. We spent some time (by we, I meant, mostly Sarah James) revising our manual and providing some additional context so that a working document can be useful to someone not familiar with all the ins-and-outs of our specific project, its history, and goals. This morning, I’m going to go through it one last time and provide a brief preface that situates this finalized manuscript in the history of our project and our field work. 

Here’s my draft of the preface:

Preface

Field manuals are living documents which not only are adapted over the life of a project to suit the needs of each field season, but are interpreted daily in the field and workspaces of a project. This document is no different.

This finalized manual from the Western Argolid Regional Project is an effort to produce an honest version of the manual that both reflects the day-to-day practices of the project as well as our regular efforts to adapt the manual to the needs of the teams and slight shifts in our methods. As a result, this is a composite document that conflates and combines any number of adjustments offered by team leaders particularly during the first two field seasons of the project. For example, we developed our site revisit procedures over the first two seasons and settled on a procedure during our time in the field. There were also adjustments made to how we documented artifacts in the project storeroom in response to requests from local officials. We have included these changes in this document to reflect our practices in the field and in artifact processing. We made these changes in consultation with our team leaders who are the co-authors of this finalized text because the both made this manual work in the field and made the text itself better.

We also added an introduction that provides some broader context for the project, its goals, and its methodology. We have also added a number of appendices that reproduce our unit form, a field guide to surface visibility and conditions, and a list of abbreviations for artifact types within the Chronotype system.

The goal of publishing this document is to preserve a record of our field practices as well as to offer a resources to other projects looking to follow similar methods in their work. In the interest in making the genealogy of field practices somewhat easier to trace through grey paper documents such as field manuals, we have released this under an open-access, by-attribution, share-alike license. This allows anyone to use freely the text of this manual, but requires that this manual be cited and any future documents based on this manual to be made available under a similar open access license.

Siteless Survey and Intensive Data Collection in an Artifact Rich Environment (15 Years Later)

It is hard to believe that my colleagues, David Pettegrew, Dimitri Nakassis, and I published “Siteless Survey and Intensive Data Collection in an Artifact Rich Environment: Case Studies from the Eastern Corinthia, Greece” in the Journal of Mediterranean Archaeology 19.1 (2006) almost 15 years ago. This article has become my most widely cited publication and, in many ways, represents a touchstone for my thinking about intensive pedestrian survey until this day.

In fact, this past week, we’ve been working on a pair of articles from the Western Argolid Regional Project. One will be a fairly conventional preliminary report with a brief methodology section. We plan to submit it to Hesperia next month. The other will be a more methodological piece that we hope to submit to the Journal of Mediterranean Archaeology before the end of the year. Both pieces are a bit challenging because they involve multiple authors and an effort to balance our desire to describe our work against an interest in providing some kind of larger analysis of our methods. Plus, there’s a pandemic which seems uniquely designed to unsettle even well thought out plans. 

The article that I’d like to see us prepare for the JMA would take our 2006 JMA article as a point of departure. It’s tempting to title our new piece ““Siteless Survey and Intensive Data Collection in an Artifact Poor Environment: Case Studies from the Western Argolid, Greece.” 

The main point of our new JMA piece could be that we’ve taken some of the lessons from the Eastern Korinthia Archaeological Survey (EKAS) that we outlined in the 2006 JMA article and applied them at scale to a rather different landscape in the Western Argolid. In particular, our survey in the Western Argolid demonstrated that applying intensive collection practices to low and moderate density scatters can unpack the complexities of artifact distribution across a landscape. This approach, while it might seem intuitive, runs counter to the traditions of site-based collection which approached the highest density places in the landscape through higher intensity collection strategies such as gridded or total collection. In this context, low density artifact scatters were often relegated to “off site” status and subjected both to less intensive collection regimes and generally mapped at a lower level of spatial resolution. 

In our 2006 JMA article, we argue that these practices tend to overlook evidence for short-term, season, or low-intensity activities in the countryside. We also argued that this approach obscures the fact that many high density consist of overlapping material from various periods which might extend in far lower densities into “off site” areas. Like a Venn diagram, then, the main area of artifact densities speaks less to the range and distribution of material at a single site either over time or from any particular period and more to the visible densities that their overlap creates. 

The main critique of the kind of rigorous, siteless approach employed by EKAS is that the intensity of this approach limited the area that we could survey. Our article recognized that the intensity of Mediterranean survey could be seen as leading to a kind of “Mediterranean Myopia” that treated surface assemblages like those produced by careful stratigraphic excavation where every sherd could be the type fossil that provides a terminus post quem for the level. While this attention is warranted in excavations, it limited the ability of survey to speak to regional issues because the scale of intensive survey projects remained limited.

WARP recognized these concerns and while many of the high-intensity siteless predecessors to WARP – namely the EKAS and PKAP, a large site survey on Cyprus – remained limited in spatial extent, WARP surveyed the majority of the 30 square kilometer area allowed by the Greek Ministry of Culture. While this is not nearly as expansive as the largest Near Eastern or North and Central American survey projects which could encompass hundreds of square kilometers, a survey that covered the majority of the territory allowed by the Greek state represented a much larger survey than the territory covered by EKAS or PKAP. Moreover, in the rugged landscape of the Western Argolid, the territory surveyed by WARP represented a topographically and historically plausible micro-region. In effect, we can propose that WARP managed to implement a highly intensive survey model in a way that responded to the historical geography of the Eastern Mediterranean and the Peloponnesus.

The article, as we now have it, includes two case studies. One examines the Roman period from the end of the Hellenistic period to Late Antiquity and shows how low density scatters shed light on the activities in the Roman landscape of the survey area. At the same time, it argues that certain periods, such as the Middle Roman period, produce less diagnostic pottery that only becomes visible under more intensive collection regimes that go beyond the typical focus on diagnostic artifacts. While the chronotype system was originally designed as both a standardized method for recording ceramics and a sampling strategy for artifact rich environments. On EKAS and PKAP, field walkers only collected one of every unique kind of artifact according to fabric, decoration, vessel type, and part of vessel (i.e. rim, base, handle, body sherd). This helped the projects manage the potential processing and storage burden associated with the collection of massive numbers of duplicate artifacts. Various experiment conducted on PKAP (and reported here) demonstrated that chronotype sampling did, in fact, preserve the functional and chronological range of artifacts in high density units, but under represented the diversity of chronotypes present. For a periods like the Middle Roman with less diagnostic artifact types susceptible to being overlooked in collection and recording, a more intensive collection regime increases the potential that we would recognize this material.

The second case study evokes the analysis of Kromna in the 2006 JMA article by examining the multiperiod scatter that constitutes the high density “site” of Panayia-Trelo in the Western Argolid. Like Kromna in the Corinthia, this site represents a series of overlapping scatters. The focus of the case study will be on the Archaic to Hellenistic period during which time the region’s relationship with Argos underwent significant change. The goal of the case study is to show that regional level analysis is not only possible from projects that prioritize higher intensity collection and spatial resolution over extent, but also requires the higher level of intensity to produce nuanced historical analysis.

Today’s blog post is just a the gentlest of sketch of what this piece needs to do to be compelling and significant. The most daunting task will be to review the scholarship published between 2006 and today to see how Mediterranean survey projects have adapted their methods to accommodate varying artifact densities. Needless to say, there’s a ton of scholarship to navigate. Stay tuned. 

Kephalari Blockhouse

I know that I’m not the first archaeologist to observe that without a field season this summer, we have theoretically more time to spend thinking carefully about our material and sites, tidying data, and preparing publications. This means, at least for me, trying to get some momentum on some lingering projects.

Two, in particular, are begging for attention. First, we have an almost complete draft of the publication of the area EF1 at Polis complete. In fact, I think we could have it ready for submission in two weeks.

More pressing at the moment, though, is a little article on the Late Roman finds from the Kephalari blockhouse in the Western Argolid. These finds were discovered in Corinth storerooms a few years ago and a group of us agreed to publish them. Of course, since that time lots of things have happened including WARP seasons, Polis stuff, a PKAP volume that’s not yet done, and The COVIDs. But this spring, the article received the ultimate motivating push: my colleague Scott Gallimore wrote up the catalogue and analysis of the finds.

So now it’s time that I do my part, which is writing up the “Discussion” section of the article. My goal is to offer a concise synthesis of 7th century settlement and rural insecurity in the northeastern Peloponnesus. It’s obviously a work in progress!

 

The assemblage from the Kephalari block house adds another small body of evidence to the increasingly complex mosaic of material from the later 6th, 7th, and early 8th century in the northeastern Peloponnesus. While the presence of material from the region’s significant urban centers, particularly Argos and Corinth, is well-known, archaeologists have only just begun to unpack and understand the situation in the countryside during these decades. The small number of excavated and well-published rural sites even in the well-studied northeastern Peloponnesus creates a particularly challenge for situating the reuse of the Kephalari blockhouse in its regional context. The growing number of stratified sequences, especially from Corinth, however, has made it increasingly possible to analyze the growing body of intensive survey data from this region from the end of antiquity. This, in turn, has offered new perspectives on a number of long-standing academic debates including changes in rural settlement patterns and urbanism, the character of the so-called “Slavic Invasions” of the late-6th century, and the presence of rural refuges such as the Andritsa cave.

Scholars have recognized that the reoccupation of rural sites, such as Pyrgouthi and the Kephalari block house appear to indicate significant investment in the adaptation of existing rural sites for reuse in the late 6th and 7th centuries. The appearance of window glass at Kephalari, for example, and the large-scale reconfiguration of the Pyrgouthi tower into a farmhouse with a courtyard suggests efforts to reoccupy these sites on a permanent basis. The evidence is less extensive from the other blockhouses and pyramids of the Argolid, but it appears that these sites were cleaned up with much of the material from earlier periods removed and the interior organization of the spaces modified with new walls and additions (Pettegrew 2006; Lord 1938; Scranton 1938).

Intensive survey has produced scatters of ceramics in the countryside that not only suggest that other Classical and Hellenistic sites experienced reoccupation in the Later Roman period, but that these sites were part of a larger reoccupation of the countryside. The site of Kastraki, for example, in the Inachos Valley, while unexcavated, may well be a similar site to Pyrgouthi or Kephalari in that it was a Classical or Hellenistic tower set atop a low rise in the valley bottom surrounded by a scatter of Late Roman material. The site of Any Vayia in the southeastern Corinthia likewise produced a low-density scatter suggesting a possible short-term reoccupation (Caraher et al. 2010) which found parallels elsewhere including on Euboea (Seifried and Parkinson 2014) and at the Vari House in Attica (Pettegrew 2006, p. 33).

Other smaller sites with material dating to the late-6th and 7th centuries exist throughout the Western Argolid survey area in the Inachos Valley and generally follow a pattern of settlement present in the 5th and 6th centuries. Athanasios Vionis and John Bintliff have argued for Late Antique Boeotia, urban and rural sites represent opposite sides of the same coin (Vionis 2017; Bintliff 2013). The persistence of sites in the countryside and even the expansion of activities into places like near coastal islands reflects the expansive use of diverse rural landscapes for agricultural purposes as well as nodes in regions and Mediterranean wide trade networks (Gregory 1984; 1995).

Urban sites continued to provide markets for rural agriculture, points of contact with larger imperial command economy, centers for manufacturing, and ecclesiastical and a certain amount of political authority. While the Finleyan concept of the “consumer city” should be laid to rest, work at Corinth (Sanders; Rothaus; Brown), Athens (Hayes), and Argos (Oikonomou-Laniado 2003) and in Boeotia (Bintliff, Vionis) have demonstrated that urban areas in Late Antiquity continued to serve as key places in Greece into the 7th century with continued investment in monumental architecture, urban amenities, and public spaces fortified in part by the growing spiritual, political, and economic role of urban bishops and the persistent reach of the imperial government.

This is not to suggest that the 7th century was not a period of significant disruption in southern Greece. Urban areas clearly experienced contraction and settlement in rural areas and this is visible in the larger WARP survey area as well as in urban surveys in Boeotia. The changes in rural settlement, including the emergence of fortified settlements in the countryside, seem to accompany continued economic activity in rural areas. While the evidence for such sites in the Argolid remains limited — the site of Kastro near the village of Tsiristra being a possible exception — the reoccupation of places like the Kephalari block house may well represent the need for both additional security and as well as continued economic viability in the countryside (Vionis 155-157). The reoccupation of fortifiable, if not necessarily fortified, sites in the Argolid may also shed light on the status of sites like the Andritsa Cave. If continued occupation of the countryside indicated the continued viability of markets and networks open to agricultural production and the fortified sites not only in Greece but across the wider Eastern Mediterranean reflects larger insecurity in the region, then places like the Andritsa Cave may well reflect the local realities of both rural wealth and instability. The so-called isles of refuge first recognized by Sinclair Hood and critiqued by Tim Gregory in the 1980s and 1990s, may also reflect the same effort to reconcile economic potential with the need for added security during unstable times.