Diagrams & Analysis

The first rule of diagramming is the hardest to obey: each diagram should communicate exactly one idea. A diagram that tries to show circulation and sunlight and structure and views all at once shows none of them, because the reader cannot tell which relationships matter. If you have four ideas to convey, make four simple diagrams, not one complicated one - a small series of clear diagrams reads faster and looks more rigorous than a single dense composite. Before you draw anything, finish this sentence: 'this diagram shows that...'. If you cannot finish it in one clause, you do not yet know what the diagram is for.

That single idea then dictates every mark. Everything that carries the idea gets emphasis; everything else recedes to a quiet base or disappears entirely. In a circulation diagram, the routes are bold and coloured and the building is a pale grey ghost behind them; in a massing diagram, the volumes are solid and everything else is gone. This figure-and-ground discipline - one thing forward, everything else back - is what makes a diagram legible in a glance. The commonest mistake is giving the base drawing as much visual weight as the idea on top of it, so the two fight and the eye cannot settle.

> Finish the sentence 'this diagram shows that...' in one clause. If you can't, you don't yet know what the diagram is for - and neither will your reader.

Name the type of diagram you are making, because each has a job: a parti or concept diagram captures the organising idea in its simplest form; a site or context analysis diagram shows what on the site shaped the design; a circulation diagram shows movement; an exploded axonometric unpacks how parts assemble; a programme diagram shows how uses are organised. Knowing which one you are making keeps it focused - a circulation diagram is not the place to also explain your structure - split them, and each one reads far faster and looks more considered.

A diagram is an act of reduction, and the test of a good one is that you cannot remove anything more without losing the meaning. So draw the idea, then take things away: remove the drop shadows, the gradients, the faux-3D bevels, the second and third colours, the arrows that do not point to anything, the labels that repeat what the picture already says. Keep reducing until the diagram is on the edge of too sparse, then add back only what the reader genuinely needs to understand it. Decoration is not clarity - a diagram dressed in effects looks busy and amateur, while a diagram reduced to clean flat shapes, a couple of line weights and one or two colours looks like the work of someone who thinks precisely.

Colour in diagrams must be systematic, not decorative. Use a small, consistent palette where each colour means something - one colour for the new intervention against grey existing fabric, one for public versus private, one accent for the single most important move - and use the same colour for the same meaning across every diagram in the project. When colour is random or purely aesthetic, the reader wastes effort trying to decode a system that is not there. A restrained palette of one or two purposeful colours on a neutral base almost always beats a rainbow, and it ties your diagrams into the visual identity of the whole portfolio.

Consistency of style across a set matters as much as any single diagram. Decide on line weights, arrow style, colour meanings, a font for labels and a level of abstraction, and hold them across every diagram in the project so the set reads as one thinking process. A quick, useful habit: draw diagrams the way you would explain the idea to a friend with a pen on a napkin - the napkin version is usually clearer than the polished software version, because it contains only what matters. Then rebuild that clarity cleanly, keeping the reduction the napkin forced.

Even a clear diagram needs a little scaffolding so the reader never has to guess. If a diagram uses colour or symbols to mean things, give it a small, quiet legend - a few swatches with one-word labels - placed unobtrusively so it explains without dominating. A legend is a courtesy that lets a reader trust the diagram; a colour-coded diagram with no key asks the reader to reverse-engineer your system, which they will not do. Keep the legend minimal, though: if you need a legend with ten entries, the diagram is trying to say too much and should be split.

Label directly and sparingly. A short label placed right on or beside the thing it names is far clearer than a numbered key that sends the eye hunting back and forth. Label only what needs naming - not every element, just the ones that carry the idea. Use a consistent, quiet type for labels (the same one across all your diagrams), sized to be readable at portfolio scale, and never let labels become paragraphs; if a diagram needs a sentence of explanation, that sentence belongs in the caption, not floating in the frame.

Finally, design the reading order. A diagram, like a page, has a path the eye should take, and you control it with size, colour, position and sequence. For a diagram that shows a transformation - a massing that pushes and pulls, a volume that splits and shifts - a small left-to-right sequence of three or four steps reads instantly, each step showing one move, arrows carrying the eye from one to the next. Sequenced diagrams like this are one of the most persuasive things in a portfolio because they show not just the result but the logic, which is exactly what a reader is assessing. Number the steps, keep each frame identical except for the one thing that changes, and let the series do the explaining.

There is a specific integrity issue with analysis diagrams worth naming plainly. An analysis diagram - of sun, wind, views, noise, movement, context - is supposed to show something that genuinely informed the design. When it does, it is powerful: it demonstrates that your design decisions came from real reasoning about a real place, which is what tutors and thoughtful employers most want to see. But analysis diagrams are frequently produced after the design is finished, as decoration - sun-path arrows and wind roses added to look rigorous, with no actual connection to any decision. Experienced readers can smell this immediately, and it damages your credibility more than having no analysis at all.

So make analysis honest by closing the loop: for every analysis diagram, show or state what the design did in response. If the sun analysis matters, the next diagram or the caption should show the shading or the orientation that resulted; if the noise analysis matters, show the buffer or the plan move it drove. An analysis with no visible consequence is decoration; an analysis paired with its design response is an argument. This is also the honest version of the storytelling you learned in Module 1 - problem leading to idea leading to resolution - expressed in diagram form.

Diagrams also serve interiors, not only buildings: a furniture-layout or circulation diagram, a zoning diagram of public and private, a lighting-layers diagram, a diagram of how a material or datum runs through a scheme - all communicate interior thinking fast, and all follow the same rules of one-idea, reduction and honesty. Whatever the scale, a good diagram is the clearest evidence that you understand your own work well enough to explain it simply - and simplicity, here, is the hardest and most convincing thing to achieve. Draw fewer, clearer, honest diagrams, and let each one hand your reader an idea in a second.