Excavation Grid and Context Sheet

In a shallow box, lay down sand, then a thin layer of dark soil, then more sand, burying different objects in each. Have someone else build it so the excavator does not know the answer.

Before the top layer goes on, dig a small pit through the layers below and fill it with a different material. That pit is a cut and its filling is a fill — two separate contexts, and the pair that teaches the whole idea.

Fix one permanent marker at a corner and call it the datum. Every measurement on the site is taken from it. If the datum moves, the entire record becomes fiction.

Run string in both directions from the datum to make squares of a fixed size — 100 mm squares suit a box, 1 m squares a real trench. Check the diagonals are equal so the grid is square, not a parallelogram.

Letter one axis and number the other, so every square has a unique reference like C4. Two coordinates plus a depth give any find a position in three dimensions.

Photograph the surface with a scale and a north arrow in frame. This is the last moment the site exists intact.

Remove the topmost deposit and nothing else, working with a trowel drawn toward you in even strokes. Layers come off in the reverse of the order they went down — that reversal is the law of superposition doing your dating for you.

Number every deposit, cut and fill in the order you meet it: 001, 002, 003. A context number is never reused and never renamed, because every other record points at it.

For each context record: number, type (deposit, cut or fill), colour, texture, inclusions, thickness, its grid extent, and what it physically touches. Write the sheet while the context is still there to look at.

On each sheet write what this context is above, what it is below, and what it cuts or is cut by. These three relations are what turn a pile of sheets into a sequence.

Everything from one context goes in one bag labelled with that context number. An unlabelled find is a souvenir; a labelled one is evidence.

Draw notable finds onto a squared plan with their grid square and depth below datum. Position within a context can matter as much as the context itself.

Leave one wall of the box undug to the end and draw it in profile with all the layers labelled by context number. The section is the visible proof of the sequence your sheets describe.

Turn the relationships into a diagram: each context a box, lines running from later down to earlier. Contexts with no relationship sit side by side. The finished matrix is the site's chronology stripped of everything else.

Read off the matrix: the fill of a pit must be younger than every layer the pit cuts through. That single deduction — which nothing in the objects themselves tells you — is what stratigraphy is for.

Hand your sheets, plans and matrix to the person who built the box and ask them to check the order you reconstructed. Anything you got wrong points at a field you left blank.

Single context recording. The system of giving every deposit, cut and fill its own number and its own sheet was developed by Edward Harris and Patrick Ottaway in 1976, from a suggestion by Laurence Keen, and was worked into a full field method by the Department of Urban Archaeology at the Museum of London. It spread from there in the mid-1980s to Scotland and York and then far more widely. It suits deep, complicated urban sites where dozens of contexts can overlap in a few square metres.

Why one sheet per context. Because the alternative — describing a trench as a whole — hides the joins. Recording each unit separately forces the excavator to decide where one deposit ends and the next begins, in the field, while the evidence is still visible. It also means two people digging different parts of a site produce records that combine without translation.

The Harris matrix. Harris's diagram shows only stratigraphic relationships — earlier, later, or no relationship — and deliberately throws away everything else. That austerity is the point: it separates the sequence, which is observed, from the interpretation, which is argued. You can disagree about what a pit was for and still agree about when it was dug.

What a context sheet cannot hold. Field recording always involves interpretation — deciding that two patches of soil are one deposit or two is a judgement, made once, by one person, and it is not recoverable later. Researchers who have gone back through archived context sheets find exactly that: the primary record contains interpretation as well as observation. Good practice is to write what you saw and what you concluded as separate statements, so a later reader can tell them apart.

The rule that follows from all of it. Excavation is the only science that destroys its own evidence as it collects it. There is no repeating the experiment. That is why the paperwork is not bureaucracy around the dig — it is the dig, and the hole in the ground is the by-product.