Stratigraphy: The Deepest Layer in a Dig Isn’t Always Oldest

How does stratigraphy work?

Steno was writing about rocks and fossils, not ruins. His Prodromus asked how one solid body came to be enclosed in another, such as a shell inside stone. Along the way it stated what can be known about the place of layers, or “Beds” in the 1671 translation, that settled out of a fluid (Exhibit A, pp. 29–30; Exhibit B, pp. 42–43).

Title page of Steno’s Prodromus, printed at Florence in 1669 and dedicated to Ferdinando II, Grand Duke of Tuscany. Its pages 29 and 30 hold the propositions on layers that this section quotes from the 1671 English translation.

Two of his propositions carry the rule. The second, in the 1671 English, says: “At what time there was form’d one of the upper Beds, the lower Bed had then already obtained a Solid consistency.” The fourth draws the consequence: “when the lowest Bed was laid, none of the upper Beds was extant” (Exhibit B, pp. 42–43). Quotations from the 1671 translation are given with the long s printed as a modern s.

Archaeologists apply the same logic to a different kind of layer: not beds that settled out of a fluid, as Steno described, but floors, rubbish and fills that people and nature left on a site. A floor laid on a layer of ash is later than the ash. A pit is later than every layer it cuts, and the soil that fills the pit is later still. In the section above, the result is a chain: topsoil 1 is the latest unit and natural subsoil 7 the earliest.

What breaks the order?

Steno knew that layers can shift after they form, and he gave two ways their position could change: “a violent excussion of the Beds upwards,” and “a spontaneous falling down of the upper Beds” when the material under them was taken away (Exhibit B, p. 44). He had fire, water and underground caverns in mind, but the same two movements, things pushed up and things sinking in, happen on a small scale on an archaeological site.

Three ways later material can end up below older layers: a pit dug from a later surface, a burrow, and a collapse into a hollow. Schematic drawings, not to scale.

The commonest break is human. A pit, grave, well or posthole is dug from one surface and goes down through older layers. Whatever falls into it, or is buried in it, now lies deeper than material that is centuries older. In the illustrative section, the sherd at 92 cm sits in pit fill 2, which is later than layers 4, 5 and 6 even though it lies below most of them.

Living things move material too. Charles Darwin wrote in 1881 that coins, ornaments and stone tools dropped on the ground “will infallibly be buried by the castings of worms in a few years” (Exhibit E, ch. 4). On his estate at Rushmore, on the Dorset and Wiltshire border, the excavator Augustus Pitt-Rivers found the burial mounds “uninjured save by the roots of trees, which were often found to have done much damage to the fragile urns” inside them (Exhibit C, p. xiii).

What can stratigraphy show?

What it can show well is which deposit came before which, and which finds were sealed together. Pitt-Rivers turned that into field rules. Publishing his excavation of a Romano-British village at Woodcuts, on Cranborne Chase, in 1887, he explained why his record was fuller than usual: an earlier dig on the same village had reported no pits at all, while he had found 95 (Exhibit C, p. xvii).

“Excavators, as a rule, record only those things which appear to them important at the time,” he wrote, so “Every detail should, therefore, be recorded” (Exhibit C, p. xvii). At Woodcuts that meant a rule about depth. Once pits turned out to be common, every object found within a foot and a half of the surface was kept apart from the rest. In his words: “All relics found within that depth have consequently been recorded as surface finds, even when they occur over a pit” (Exhibit C, p. 12).

Pottery exposed in place beside a scale bar during an excavation in Israel, August 1957, photographed by Benno Rothenberg. Recording finds where they lie, before they are lifted, is what lets the order be checked later.

The point of the rule is that a find logged simply as a pit find could have come from the topsoil above the pit. Keeping the two apart costs nothing at the time and can save a wrong date later. The Egyptologist Flinders Petrie made the same argument from the other side in 1904. Groups of objects burnt in pits and then covered over, he wrote, “cannot possibly have been disturbed later on to insert objects,” because the ash around them was undisturbed (Exhibit D, p. 145).

Petrie also compared archaeological proof with the evidence used in court, insisting that “there is not one logic for the present, and a different logic for the past” (Exhibit D, pp. 136–137). This site’s retrial method uses that comparison throughout. The same way of reading the ground, for what it can and cannot hold, is applied to Proctor’s Ledge, where Salem’s hanging site was traced from deeds, family memory and the lie of the land.

How is the order written down?

On a large site, hundreds of units touch one another, and many contacts say nothing new. If topsoil 1 lies on pit fill 2, and the pit was cut through ash layer 4, then 1 is later than 4 whether or not the two touch anywhere. Archaeologists draw the order as a diagram, latest at the top, that keeps only the contacts needed to fix the sequence. It is called a Harris matrix, after Edward C. Harris, whose Principles of Archaeological Stratigraphy appeared in 1979 (Exhibit G).

The illustrative section as a sequence diagram, latest unit at the top. Solid boxes are deposits; dashed boxes are cuts. Drawn from the table published with this page.

The section has 13 observed contacts between its nine units. Four of them, such as topsoil 1 lying on ash layer 4, follow from the others and drop out, leaving 9 lines in the diagram. The diagram also shows what the section cannot decide. The pit (units 2 and 3) and the posthole (units 8 and 9) never touch, so the order between them is unknown: 4 pairs of units have no order at all.

What can’t a layer tell you?

A layer has an order, not an age. Nothing in the section says whether layer 5 took a week or three centuries to build up. Even its contents can mislead, as Petrie warned: “in strata or tomb-groups there is a chance of older things being reused” (Exhibit D, p. 139). An old coin or pot can end up in a much later layer.

Illustration

An illustration, not a record: excavators of the 1890s at a trench wall. The bands show the order in which layers formed; nothing in them shows how many years each took.

Dates come from outside the layers. They come from objects that can be dated, such as coins or inscribed pieces, and from laboratory methods such as radiocarbon, which this page does not cover. Stratigraphy then spreads those dates through the sequence. A coin sealed under a floor gives the earliest date the floor can have, the logic set out on the page on terminus post quem, the newest part setting the earliest date.

Layers out of their expected order came up in a Tennessee courtroom in 1925. Questioned by Clarence Darrow on July 20, William Jennings Bryan said that George M. Price, whom he described as a professor of geology, pointed out that “attempts to measure age by those layers where they are not in the order in which they are usually found makes it difficult to fix the exact age” (Exhibit F, p. 297). The exchange is retried on the Scopes trial page, where Darrow put Bryan on the stand. On an archaeological site, out-of-order material is not a reason to give up on order; it is a sign to look for the pit, burrow or collapse that explains it.

A worked case: Petrie’s town mounds

Petrie showed in 1904 how tempting it is to read depth as time, and how far to trust it. For Egyptian town mounds, he suggested allowing “a rate of growth of 20 inches a century,” with Nile mud adding about 4 inches a century in cultivated ground (Exhibit D, pp. 10–11).

Petrie’s two estimates for a Delta mound whose deposits end about 500 A.D., in years of build-up. Drawn from the table published with this page, from Petrie (1904).

Petrie’s town-mound arithmetic, with page references (CSV) (where the data comes from) Data: W. M. Flinders Petrie, Methods & Aims in Archaeology (London: Macmillan, 1904)

His example was a Delta mound about 40 feet high whose deposits end about 500 A.D. Forty feet is 480 inches; at 20 inches a century that is 24 centuries, or 2,400 years of build-up, which Petrie rounded to “say 2000 B.C.” Allowing for a buried base 5 feet lower and a net rate of 16 inches a century, 45 feet gives 33.75 centuries, or 3,375 years, which he gave as “about 2900 B.C.”

The two figures for the same mound differ by 975 years, and the only change is in the assumptions. Petrie did not hide that: “No exact conclusion could be based on this; but it is a valuable clue” (Exhibit D, p. 11). The order of the layers in such a mound is evidence. The number of years is an estimate that depends on a rate assumed for that mound, not measured on it.