More college football stats…

So, now that the college football regular season is over, I thought I’d take a look at how the teams did this relative to the talent they recruited.


So, team talent predicts more than half of the variability (but it’s worth noting that I’m running a rank against a rank, which makes regressions a little screwy since the values can’t dip below zero).

I would use this regression to try to predict a season for betting purposes. I tried (but didn’t bet), and the results were hilariously bad.

Why? Coaching.

The closer the dot is to the top right corner, the better the team did above and beyond what you’d expect from talent alone. The closer the dot is to the bottom left, the more the team under-performed.

So, take Tennessee for example. Based off talent alone, this equation would have predicted that they should have finished ranked #21 in the polls. The team that finished #21 was Michigan with an 8-4 record. That’s where most people predicted they’d be, and I guessed 9-3 based on looking at the schedule and seeing how many teams were more talented than they were. I was flabbergasted when a friend of a friend of mine predicted they’d be 4-8.

Instead, they finished #92 with a record of 4-8. This record made Tennessee one of the top 10 under-performers in all of college football.

Moreover, if you look at the teams along the top, right edge of the pattern, and you see a bunch of teams that have coaches that either got contract extensions or left for better jobs.

Along the bottom, left edge – a murderer’s row of schools who fired coaches or have coaches on the hottest of hot seats.

Edit: here’s some updated graphics after cleaning up the data a little more, excluding FCS schools, and using Sagararin Ratings versus 24/7 Team Talent Composite points.






College Football Stats…

You know, I’ve tried for awhile to come up with a way to predict how well my college football teams (Tennessee, Arizona, and Mississippi State) are going to do prior to the start of the season. One of the best predictors for the previous season is the 24/7 talent composite, and this year’s version was released on Thursday. However, it’s not a strong enough predictor that you’d want to take it to Vegas.

Regardless, I put it in graph form with the 24/7 composite rank on the Y-axis, and the percentage of each roster that is composed of recruits that were ranked as 4 and 5 stars coming out of high school on the x-axis.


Obviously, the correlate. However, relationship of the continuous variables relative to the rankings helps show, for example, how far ahead Bama is to everyone else.

Not surprisingly, talent is distributed non-linearly, and as fans of SEC football, you know that the rich get richer (ahem… f@#$ing Alabama).
You can also see a gap between Alabama and the next cluster of teams (Georgia, LSU, and Auburn) and then Tennessee, Florida, and Ole Miss.

However, when you look at ranking versus talent average (i.e. the average number of “stars” each player was ranked when coming out of high school), those breaks smooth out.


So, looking at this, I’m guessing 9-3 for Tennessee just based off talent alone.

Geophysics at Hester

This summer’s AMEC Excavation field school is at the Hester Site, outside of Amory, MS. It’s a pretty well-known Paleoindian site in the southeastern United States that was excavated by Sam Brookes in the 1970s. Our excavations this summer will form the basis of Jim Strawn’s master’s thesis.

You can read the report here, but what makes it unique is that is has stratified Paleoindian and Early Archaic deposits. He’s the profile from the original report.

Screen Shot 2017-05-13 at 6.58.59 PM.png

The tough part for us right now is determining where exactly the original excavations were situated. For starters, there’s a sign…

Screen Shot 2017-05-13 at 7.04.52 PM.png

Hunter Saunders. Sign model.

However, we’d like a little tighter spatial resolution than that. So, I downloaded some LIDAR data and used it to “wiggle match” the contours from the original map to the contours from the LIDAR data. Luckily, there’s a spot where it looks like a front-end loader or something scooped out some ground in the immediate vicinity, and that spot is present on both the original site map and it pops up with the LIDAR-generated contours.

Hester Lidar_Burress_Sketch.jpg

Original site map geo-referenced to LIDAR data.

To actually ground truth this, we asked Tony Boudreaux and Stephen Harris from Ole Miss to do some geophysics to see if we could find the location of the excavation units using a gradiometer and ground penetrating radar.

A little geophysics action at Hester…

A post shared by Shane Miller (@the_durty_trowel) on

Luckily, based on the gradiometer results, they think they found the location of the excavation block. It probably helps that the remains of an old shack were thrown into the excavation block when it was originally backfilled. Stephen and Tony also think they that have a signal from where the shack originally stood, as well. So, based on their results, here’s where I think the excavation block from the 1970s was.


They were also able go get in an east-west transect with the GPR across the midline of gradiometer grid , which is difficult because all of the dense vegetation. Based on this, they think they found where the trench is located, which lines up with where the trench should be if what we think is the excavation is actually the excavation block.


Yes, I do owe those folks from Ole Miss some cold beverages.

Swag Update – Spring Break 2017

Since I just finished my SAA paper, I have a little time to post a quick update from this past Spring Break. Derek and I took a few students out to try to finish up the last couple of units that were “in progress” when Tropical Storm Bonnie hit us last spring.

We spent the week digging. And plotting. And digging. And we’re still not through it all. This is hands-down the densest excavation block at the Allendale Quarries. Here’s the GoPro video.


Also, last year I put in a series of bucket auger tests, and in one of them south of the Swag excavation block, I turned up a bunch of really weathered flakes. So, we got DuVal Lawrence, Leslie Page, and Will Joseph to hammer out a test unit. About 50cm below the surface, they turned up a layer of artifacts that contained…drumroll…

Swag Taylor Point

That would be a Late Paleo/Early Archaic “Taylor” point (the artistic handiwork was done by Hunter Saunders).

Even better…there’s stuff below it, too.

Field School Update -Analyzing All of the Rocks

One of the downsides of analyzing assemblages from quarry sites is obvious – they generate a lot of stuff.

Thankfully, I had some funds left over in the field school budget to hire two students (Cody Oscarson and Hunter Saunders) to analyze the assemblage.


Hunter Saunders and Cody Oscarson

First, thanks to Jesse Tune and the field school students, we managed to get almost all of the “screen bags” sorted, size-graded, counted, and weighed. Cody finished up the last few bags and then entered them into our Access database.


Access Database

Then, Cody and Hunter started working on analyzing the piece-plots, or the artifacts for which we have 3D coordinates. Each artifact gets a recording slip that they fill out with some basic info (size grade, weight, presence or absence of cortex) and their interpretation of what the artifact is (flake, biface, uniface, core, etc.).


Piece Plots and Recording Slips

Finally, while Cody has been tackling the bulk of the lithic analysis, Hunter has been figuring out how to draw artifacts. He’s getting pretty good at it, while also learning his way around Adobe Illustrator.


An artifact drawing in progress…


…and one of his pieces that’s further along. 

So, with the help of these guys, and Ryan Young working on the particle size analysis at the UT-ARL, we should have all of our ducks in a row to put together our poster for SEAC.

Field School Update – A quick and dirty comparison of Topper Clovis to the Swag Site

While we were excavating at Swag this field season, I kept saying over and over again to Derek Anderson that the stuff we were finding was eerily reminiscent of the Clovis assemblage at the Topper Site – in particular the part of the Topper site that I analyzed as part of my master’s thesis.

I’ll just straight up admit that it kind of weirded me out, especially since we excavated my master’s thesis block exactly a decade ago. (Typing that doesn’t make me feel old at all).

So, what made my thesis block unique at Topper? Well, allow me to recap with a trip down memory lane… I finished my thesis in 2007, moved to Arizona immediately after, and then May 2008 I went back to Topper to do more fieldwork, because I like to party. Topper in early May is particularly notorious for some really bad thunderstorms, and the best place to hole up to wait for them to pass is the cinder block bathroom/shower house. While we were waiting to the storm to pass, Ashley Smallwood and I got into an argument about the nature of the Clovis assemblage in the area of the site she was writing up as part of her dissertation versus the part of the site I analyzed for my thesis. I thought the stone tools were pretty big, rough, and more early stage rejects. After looking at her stuff, Ashley thought the assemblage from her area was much more refined, or much closer to a “finished product” if you will.

Another way of putting it is that if you think of lithic production like an assembly line. I thought our stuff was more of what you’d expect at the start of the line, and she thought our stuff was much more of what you’d expect at the end. After running some numbers, we discovered that despite the close proximity of our excavation blocks,  we had very different assemblages. Since my block was closest to the chert source, I had the early stage stuff, and as you moved uphill to Ashley’s block the artifacts were closer to the finished product.


Then, to make it even more interesting, Doug Sain analyzed a third block and found that the composition of his block was different from mine and Ashley’s blocks. Whereas our blocks were mostly rejects from biface production, Doug’s was mostly debris from producing prismatic blades.

In other words, if Ashley and I had two ends of an assembly line, Doug’s block was like stumbling onto a different assembly line. Below is an image from a book chapter we co-authored that sums up what we think it all means…


Back to my thesis block, in addition to having bifaces that were in the early stages of production, there was a lot of large amorphous cores, big flakes, and a lot of large tools clustered at the end of the block closest to the chert outcrop. One of them (the top left of the picture below), had ground margins (like what you’d expect to find if it was hafted) and the business end was beat all to hell (like it repeatedly bashed against something really hard). Goodyear and I argue that this artifact and a few other large tools like the one in the bottom left of the picture below might be indicative of heavy duty woodworking going down at the site. However, once you move away from the chert source, the artifact density diminishes and you get more formal tools like the scrapers on the right, including the artifact in the bottom right that is made from raw material from North Carolina. 


There was a large blade tool, expedient scrapers, and something approaching a formal blade core. There’s also several artifacts with denticulated (i.e. sawtooth) edges.


Screen Shot 2016-07-09 at 5.34.35 PM

Formal, hoof-shaped blade core in the center.

Screen Shot 2016-07-09 at 5.39.15 PM

Both have edges that are modified, and the blade on the left appears “backed” (i.e. the edge on the non-business side was intentionally blunted).

Screen Shot 2016-07-09 at 5.59.48 PM.png

Denticulated Scrapers

And we also had a few temporal smoking guns, like overshot flakes…Screen Shot 2016-07-09 at 5.35.17 PM.png

And the really, really white biface in the bottom center. It’s the medial portion of a preform that been fluted on both sides.

Screen Shot 2016-07-09 at 5.35.46 PM.png

So, where do I see similarities between my thesis block and what we dug up at Swag this summer?

Both have REALLY large flakes and cores and a LOT of them.

Both have blade cores that are “hoof-shaped.”

Both have large, heavy duty scraper/adze tools.

Both have modified blade tools, including several with denticulated edges.

Both have a lot of expediently modified flakes (i.e. things that were potentially modified and used for the task at hand, and then discarded).

The biface assemblages of both blocks are pretty clunky and skewed heavily towards early stage reduction.

I don’t have a picture of it yet, but Jesse Tune also pulled a flake made of raw material that might also be from ryholite from North Carolina (like the scraper from my block).

What’s missing? Unfortunately, the temporal smoking gun – overshot flakes and/or bifaces that are far enough along in reduction that we can confidently say what “type” they are. At Swag, however, there are over-face flakes (i.e. a flake that went passed the midline of a biface, but didn’t go all the way across to take a bite out of the the opposing edge).

However, Derek and I are thinking that we just might be smack in the middle of a early stage reduction (and heavy duty woodworking?) area like my block at Topper appears to be. If that’s the case, then if it’s anything like Topper, perhaps we should expect to some areas nearby that look like Doug’s block (lots of blades) or Ashley’s block at Topper (smaller debitage, later stage biface reduction, more tools)? If we can find a spot like that, that’d be pretty sweet, because that’s where we’re likely to find a biface that would let us know how old our assemblage is, or perhaps a hearth that will let us get a radiocarbon date.

With that, I leave you with this throwback video of my students working in the rain while listening to Biggie Smalls, because my man Robert Martinez showed up to field school with an iPod locked and loaded with a wide array of jams.

Field School Update – Some artifacts

While we were finishing up with fieldwork, several of us got an email from Metin Eren. For those of you who don’t follow Paleoindian archaeology closely, Metin is a publishing beast. So, of course, he was sending us an article he co-authored with Briggs Buchanan, another publishing beast.

As soon as I opened it, I thought, “Well, this is timely.” It’s an overview of Clovis lithich technology.

There’s a lot of things I find interesting about these two writing a paper together on this topic. First, I would consider Buchanan to take more of a hard “Clovis-first” position (i.e. there is no pre-Clovis at all) and Metin more of a soft “Clovis-first” position (i.e. there’s solid evidence for pre-Clovis, although Clovis was first in some areas like the Great Lakes). Now, I could have this wrong, because people change their minds, but that’s how I read their thoughts from their extensive publication record.

Second, for the most part, they both come at studying stone tools from two different directions. Metin is an amazing flint-knapper, and he has written a ton on the variation in how people produce stone tools. You can almost think of Metin as someone who has spent a lot of time reverse-engineering the steps in the process of making a stone tool. Briggs, on the other hand, has written extensively on applying “cultural transmission theory” to the finished products. He is interested in what the slight variability in the shape of items like projectile points can tell us about prehistoric learning networks. Granted, there’s a lot of overlap between the approaches and the questions they’re trying to answer, but they’re just different enough for me to be intrigued about how they view Clovis technology.

So, what did they come up with? Well, this baller graphic for one:


Why do I think this is timely? Well, at Swag we have a lot of artifacts, and the overall character of the assemblage seems to fit really well into this lithic tech “cookbook.” Moreover, this flowchart REALLY gets at the heart of the Clovis assemblage at Topper, but Topper is a great example of how this gameplan gets adapted to poor quality raw material. Also, the assemblage at Swag bears some striking similarities to the Topper Clovis assemblage, and in particular to the artifacts that came out of the excavation block that formed the basis for my master’s thesis.

Over the course of the excavations, I noted some of the artifacts that illustrate this to some degree, and I pulled them out and took some photos last week. I’ll also compare them to the Eren/Buchanan flowchart. In a later post, I might compare these to the artifacts from the Clovis assemblage at Topper.

First, it seems like they were hauling in some some big chunks and spalls directly from the outcrop, like this beast of a flake.


And from there, we have have a lot more opportunistic knapping than prismatic knapping. However, we do have a few attempts at creating the blade cores for prismatic knapping. However, they’re not very pretty compared to Clovis sites like Gault or Carson-Conn-Short, which are sitting in close proximity to really abundant, high quality material. Rather than having true, formal conical- or wedge-shaped cores, we have this…


You can see where they peeled off a lot of long removals to create long,skinny flakes, but they they were also trying to overcome some impurities in the raw materials. I’ll save you the suspense and just say that the other side of this piece is ugly. Also, based on some of the other flakes from this unit, I suspect that when Derek Anderson goes nuts refitting this, we’ll find that they kept trying over and over again to rejuvenate this one side to of the core to produce blades.

Ok, what were they doing with those blades? Well, they were making some stuff like this…


…a long, but thick flake that has margins that have been edited to make them more jagged (i.e. serrated or denticulated).

In some cases, they took advantage of opportunistically knapped flakes to make artifacts that have the same basic shape (but were probably not produced by a removal from a formal blade core).


In other cases, they used “opportunistic flaking” to produce artifacts that look like side-scrapers (apologies for the blurry pic – I was in a hurry)…


…and larger, more heavy duty unifaces (similar to those found at Topper).


We also have several heavy duty adze-like tools that are bifacial…



…some bifacial pieces that are just, for lack of better words, clunky…


…and finally a biface that might be a bifacial preform with overshot removals (0r maybe another adze?), like the biface in the far left of the Eren/Buchanan graphic. However, their bifacial preform was placed in a cache, whereas ours was likely a manufacturing reject at a quarry. Predictably, the Swag example here is much less aesthetically pleasing. It still has some of the silt/clay mush adhering to it, and I didn’t have time to really gently remove it before taking the picture.


So, what do I think? I wish we had uncovered a fluted preform, or a biface with more clear overshot removals (like the artifacts on the far left and bottom left of the Eren/Buchanan graphic), but as it stands now, we think think have an artifact assemblage that bears an uncanny resemblance to the Clovis assemblage at Topper and elsewhere in North America.

Field School Update – Dots on a map…

So, now the real fun begins – how do you make sense out of all of the boxes of rocks and dirt?

Thanks to Joan Plummer – “Our Lady of the Database” – we were able to get back to Starkville with all of our forms entered into a database. Also, thanks to Jesse Tune donating his time to run a field lab, we have everything washed and sorted, and almost everything size-graded, counted, and weighed. Call it a blessing in disguise, but thanks to Tropical Storm Bonnie, we were able to steal a couple days for some extra field lab time that allowed us to get a lot more done than we expected.

I imagine there’s a lot of variation in how archaeologists tackle organization, analysis, and eventually writing up an assemblage. My new personal favorite way to deal with it is to say, “<graduate student name>, meet your thesis.” However, all of my graduate students currently have thesis projects, which means that this summer’s fieldwork is going to be a group effort.

Since Derek is already down in St. Croix, Jesse in Costa Rica, Kelsey working on her thesis, and Ryan at Russell Cave, I figured I’d start chipping away at it first.

I see archaeological sites as first being a database and spatial analysis problem to solve, and that’s not a new thing. In fact, if you got back and look at some of the WPA collections from the Mid-South from the 1930s, they basically did an analog equivalent of GIS. They created “on paper” relational databases that they could easily connect back to maps and profiles, which has allowed folks like Thad Bissett to completely reconstruct what they did in 3D.

So, step #1 – dump all of the points we took during the field season into ArcGIS.

Swag Map

We recorded 1782 positions with the total station. Of these, 389 were “control points” used to either mark closing elevations (each corner + center) or the perimeter and closing elevations for features. The remaining 1393 points were artifacts or sediment samples taken from underneath large artifacts (>15cm in diameter).

Alright, let’s do some quick and dirty explanations of the distribution…


We managed to completely excavate one unit, and for most of the others we think we were past the bulk of the artifacts-bearing deposits. There were two units that we were above the densest levels. Then, we had one unit that shattered the all time record for number of plotted artifacts for the Allendale Quarry sites, which was ~80 artifacts from a 5cm level. One 5cm level from our densest unit produced 329 plotted artifacts. That’s pretty nuts.

We have three empty spots on the map that are easily accounted for: 1) Our test unit from 2015, 2) a tree throw , and 3) a burnt tree stump. However, if you take into account that not all of the units were completely excavated, Derek Anderson noticed in the field that there seems to be something of a trend with our excavation block…


The southernmost unit is by far the densest, and that diminishes as you go north, and probably west.

Next post, I’ll discuss some of the artifacts that we’ve found, and our rationale for why we think we probably have a pretty old site on our hands.

A post shared by Shane Miller (@the_durty_trowel) on

Field School Update – Week 4

Remember this from the end of last week?



Well, after surviving Tropical Storm Bonnie, we arrived at our excavation block to find it…

Tropical Storm Bonnie rocked us.

A post shared by Shane Miller (@the_durty_trowel) on

…totally wrecked.

For awhile now, Derek and I have been wondering, “Aside from the raw material for making stone tools, why were people here of all places?” We have a hunch that, since it’s down in a bowl-like depression, the Swag site might have been in a swamp or bog like setting – not unlike the Carolina Bays that dot the landscape of the surrounding area.

Maybe it looked something like this (I shamelessy stole this photo from Andy White’s blog recounting some work conducted by Chris Moore on Carolina Bay formation)…


So, rather than looking like the partially logged pine forest today, Swag may have at one point looked like this:


But, adjacent to our “Swag Bowl” is an old oxbow of the Savannah River that dates to before the Last Glacial Maximum (LGM), sometime around 25,000 years give or take a few Thursdays.

So, my best guess on what happened based on the landscape setting and the character of our sediments…

  1. The “Swag Bowl” may have been entirely closed at one point in time and behaved a lot like the Carolina Bays found in the immediate vicinity.
  2. The Savannah River bumped into it, opened up the west side of the bowl, and provided a channel for the bowl to drain.
  3. This channel incised, exposing the chert outcrop on the west side of the bowl.
  4. The interior of the bowl went from being a bog to grassy. In other words, it was dry enough to be grassy, but too wet for most trees.
  5. Clovis folks rolled in and found a flat, grassy spot with some pretty nice chert a few dozen yards away. It may have also been surrounded by oak/hickory forests on the high ground. Not too bad of a place if you’re a hunter-gatherer.

Well, we think they’re Clovis folks. I’ll pull some artifacts and photos and explain why we think that, and what we plan to do next, in a future blog entry.

So, what did this mean for our attempt dig square holes in the bottom of this bowl? It meant that the water filled the plastic that covered our units was the least of our worries…

Pumping down the water in our excavation block…

A post shared by Shane Miller (@the_durty_trowel) on

…because the 4-6 inches of rain we got over the weekend all flowed to the bottom of the bowl. This caused the water table to rise dramatically. Pro-tip: A small fish pond pump hooked up to a Goal Zero Yeti 150 can move a lot of water in a short amount of time. However, it wasn’t enough. The sediment in our block turned into goop, and after much consternation (and admitting to Ryan Young that he was right about the water table), we decided to map the remaining artifacts that we could see, and then call it quits for this field season.

The upshot? We had more time to spend in the field lab. Thanks to Jesse Tune and the students, we have almost everything washed, sorted, size-graded, counted, and weighed.

We also were able to fit in a visit to see Matt Sanger‘s crew at Sea Pines (and lunch and an afternoon at the beach).

We went to see Matt Sanger guy and a shell ring.

A post shared by Shane Miller (@the_durty_trowel) on


We also made it back in time for one more sunset on the Savannah River.

The last WWP sunset of the 2016 field season.

A post shared by Shane Miller (@the_durty_trowel) on


Here’s the GoPro video from this last week…