[Draft written for review, edit before publishing]
A measured drawing is a drawing of a building as it actually is, taken from the building itself rather than from anybody’s plans. It sounds like a clerical task. It is the single most useful thing I have been taught, because you cannot do it inattentively — the building stops being a picture about four minutes in and becomes a set of facts, several of which contradict each other.
What the drawing is for
Three things, usually. It is a record, in case the building is altered or lost. It is the base for any intervention, because you cannot design into a space you have not measured. And it is evidence: a plan showing that a wall is 530mm thick at the ground floor and 390mm at the second tells you something about how the building was built and in what order, and that argument can only be made from measurements.
Sketch before you measure anything
The first pass is a freehand survey sketch, roughly to proportion, large, in pencil, with plenty of white space. Not to scale: if it is to scale you will trust it, and it is wrong. Its job is to be a container for numbers. Walk the whole building first and draw every room, every opening, every change of level and every oddity you notice, then decide where dimensions need to go.
Leave it generously spaced, because you will write more on it than you expect: material notes, the direction floorboards run, where plaster sounds hollow, which way a door swings, the fact that one window has four lights and its neighbour has six. I learned to put a north arrow on every sheet after an evening spent unable to work out which elevation I had drawn.
The kit, and what each tool is for
- A 5m steel tape for anything you can reach, and a 30m tape for overall dimensions. The long tape is what keeps your errors from compounding.
- A laser distance meter: faster than a tape, but it needs a surface square to the beam and will measure the wrong thing with great precision.
- A plumb bob and a short spirit level, which answer the question a tape cannot: is this vertical, and is this flat?
- A profile gauge for mouldings, and a scale rule for drafting up.
- A camera, with a rule in the frame for anything you will want to measure later from the photograph.
Running dimensions, not separate ones
This is the technique that changed my drawings most. The instinct is to measure each thing separately: window 1,120; pier 840; window 1,120; and so on along the wall. Every one of those readings carries a small error, and the errors add up, so by the end of a long elevation you are out by an amount you cannot locate.
Instead, lay the long tape along the wall, fix the zero at one corner and leave it there, and record every feature as a cumulative distance from that single zero: 0, 310, 1,430, 2,270, 3,390. Now each number is independent of the others and an error in one does not move anything else. Take the overall dimension last, as a check — if the sum of your intervals does not match it, something is wrong while you are still standing in the room.
Measure from one zero and take the overall last. The check has to happen while you can still reach the wall.
Triangulation, because nothing is square
Rooms in old buildings are not rectangles. They were built by eye against walls that had already settled, and then they settled further. If you measure four walls and assume ninety degrees at every corner, your plan will not close, and the error will land wherever you happen to stop drawing.
The fix is triangulation. Measure the two diagonals of each room as well as the four sides; a triangle with three known sides has only one shape, so the diagonals fix the geometry and tell you exactly how far from square the room is. For an irregular wall, run a straight baseline across the room with the long tape and measure offsets from the baseline to the wall at regular intervals. You end up with a wall described as small deviations from a known straight line, which is far easier to draft.
Heights and one datum
Floors slope, so “floor to ceiling” is not a single number and measuring heights from the floor gives you a building that does not stack. Establish one datum instead: a horizontal line marked around the building at a convenient height, traditionally a metre above the finished floor at a chosen point, transferred with a level. Every vertical dimension is then taken from the datum, up or down, and your sections stack correctly even where the floor does not.
Record what is there, not what was intended
The discipline is to draw what you measured even when it looks like a mistake. Four of the six bays are 1,120 and two are 1,090: draw them that way. The temptation to average, round or regularise is strong, and it is the main way a measured drawing becomes useless, because the irregularity is usually the evidence.
The convention that makes this liveable is to separate the record from the interpretation. Draw as-found conditions solidly; show anything you are inferring — a wall you could not reach, a sill you only saw from the street — in a lighter line, and say so in a note. The Historic American Buildings Survey guidelines, a good free model for how to lay a set out, are strict about that distinction, and it is worth borrowing even for a studio project. Measured: Beacon Hill Doorways is the set where I finally stopped tidying things up, and it is the first drawing I made that taught me something I had not already decided.
Drafting it up
Draft from the field sketch while the building is still fresh, ideally the same day, and keep the sketch — it is part of the record and the only place some of your notes exist. Draw at a scale that can carry the information: 1:50 for plans and sections of a whole small building, 1:20 or 1:10 for a doorway or a window, 1:1 for a moulding profile.
Then add what makes the drawing usable by somebody else: a scale bar as well as a stated scale, a north point, the date of survey, who surveyed it and with what, and a note of anything inaccessible. A measured drawing without a date is a drawing of a building that no longer necessarily exists in that form.
What a scanner does not do
Laser scanning and photogrammetry produce point clouds of extraordinary density, and for a complex surface — a vault, a carved capital, a settled wall — they are plainly better than a tape. But a point cloud is a measurement, not a drawing. Somebody still has to decide which of those millions of points is the wall, which is the plaster added over it, and which is a radiator. Working on the object survey for Conserving the Cast Collection made that obvious: the scan told us the shape, and only handling the casts told us what the shape was made of and where it had been repaired.
Which is the real argument for measuring by hand, at least once per building. The tape makes you touch every surface you record, and you notice things with your hands that you do not notice with your eyes.
