Building a Pocket-Sized Soil Probe for Archeology: The Friction Cone Penetrometer

Summary

From 2013 to 2016 the National Park Service's Southeast Archeological Center (SEAC) worked with Florida State University (FSU) to build a portable friction cone penetrometer (FCP): a thin probe that would tell archeological midden soils from natural ones by how the soil resists and grips a rod pushed through it. The National Center for Preservation Technology and Training (NCPTT) granted $25,000; FSU added unmeasured funds of its own.

Result: three prototypes, none of them workable. The idea — telling soils apart by their resistance and cohesion — still looks theoretically sound. Shrinking a proven, truck-mounted hydraulic tool into a light mechanical one proved too hard on the budget.

The off-the-shelf dynamic cone penetrometer the project set out to adapt (Vertex). NPS.

Why build it

Probing is old: archeologists have long pushed metal rods into the ground and felt for pots, burials or buried walls (Clarence B. Moore did it in Florida in 1918). But it is subjective — one prober feels shell where another feels nothing. Michael Russo, who had probed shell middens with rods up to 4 meters long for over 20 years, found his colleagues could not reproduce his readings.

Cone penetrometers (CPs), used in road building and farming, give an objective number — the cone index, the force needed to push a cone through soil. A friction cone penetrometer adds a sleeve behind the cone that measures the soil's friction. Fine-grained soils such as clay grip a rod more than coarse sand, so the two readings together describe what kind of soil the probe is passing through. Commercial FCPs exist, but they ride on large trucks and use a probe over 3 inches (8 cm) wide.

A portable version, with a probe under ½ inch (1.3 cm), could:

• map midden sites without shovel tests, which are typically 12 inches (30 cm) across and bring up artifacts;
• ease the curation crisis: surveys required by the National Historic Preservation Act (Sections 106 and 110) have filled museums and storage with artifacts awaiting costly analysis. Probe readings would stand in for many of them;
• reach coastal and riverside sites in saturated ground, where excavation is rarely required and ground-penetrating radar, magnetic and LiDAR methods work poorly — sites now being lost to rising seas.

Design choices

• Dynamic, not static. A static CP is pushed at a steady rate, and critics say no operator can do that consistently. A dynamic CP drops a set weight from a set height, so the force is known. Russo chose dynamic, for repeatability.
• Organic middens, not shell. Shell was the original inspiration, but repeated impacts might break the fragile friction sleeve, and breaking shell would need heavy weights. The target became organic midden soils.
• Test soils. Retired soil scientist Lynn Coultas (Florida A&M University) analyzed seven samples of about five gallons each: midden soil from the Byrd Hammock site near Tallahassee and six other soils. The midden soil differed significantly from all the others in grain size, organic matter, water and sand content.

Three prototypes

Team 16's prototype (top) and Beckers' redesign (bottom).

What the tests showed

Pace's lab tests suggested the idea works: cohesion coefficients could tell midden soils from the other soils. But:

• the resistance and cohesion signals peaked out of step;
• fixing that made the probe need ten times as many drops to move a short distance;
• he could not repeat his first good readings;
• background noise — probably the weight bouncing microscopically on its platform, like a basketball settling after a drop — blurred the peak forces the calculation depends on.

One drop in sand: the cone (blue) spikes and rings; the friction sleeve (red) lags behind it.

Why it did not work

Portability defeated the dynamic design. One person would have to carry a 10- or 25-pound weight, the long rod, recording equipment and perhaps a stand to keep it vertical.

Speed. To beat shovel testing — about 25 holes a day for an efficient archeologist — the probe needed to be at least twice as fast: under ten minutes a reading, nearly 50 a day. In the lab, setting up took about 20 minutes; repeat readings in one spot might take a couple of minutes, but moving to a new spot meant packing up and setting up again, perhaps 30 minutes each time. Archeologists could dig shovel tests faster.

The physics of impact (Shih's analysis):

• an impact lasts milliseconds, so it needs sampling in the kHz range and advanced data systems — beyond the project's time and money;
• the rebound makes the probe oscillate, contaminating the data;
• impact force responds to soil rigidity differently from friction, so without characterizing soil dynamics the results may not be reliable;
• a load cell strong enough to survive the impact loses resolution.

A way forward

Shih suggests going back to a static probe — but driven by a crank turned at a steady rate (say, one turn every 5 seconds) or a speed-controlled motor, rather than by muscle. The crank would also give the depth, doing away with the laser range finder. Gear sizes or motor power would have to be worked out for midden and surrounding soils.

The project's most useful findings, the authors conclude, are that the dynamic model is a poor basis for a portable FCP, and that Pace's cohesion measurements suggest middens can be told apart by resistance and cohesion. Future engineers may build on both.

Sources

Based on Michael Russo (NPS Southeast Archeological Center) and Chiang Shih (Florida State University), "Final Report on the Friction Cone Penetrometer," National Park Service; published by the National Park Service. The report names Shih's colleague as "Shin" and the static model as "CSP" in one passage; both are corrected here. Figure 1's manufacturer's parts drawing and three computer-drawn exploded views are left out; the student teams' own reports are its appendices.