Fig 1The goniometer benchDrag to run the year21 Jun · sun 73° up, 17° off vertical · 501 W/m² · 14.8 h
The vernier blade: it reads the lamp's height on the railThe graduated clinometer, from the horizon round to the zenithThe patch of light on the tile: its length is the whole storyThe rail is an arc struck about the tile, so the distance never movesA goniometer bench, in the northern hemisphere at 40° N. The Sun's rays are a collimated beam from a lamp on a rigid rail — an arc struck about the collector's centre, so the lamp's distance from the tile cannot change. The beam always falls on level ground: what the season changes is the angle it strikes at, and how long it stays there. At midsummer it comes in at 16.6° off vertical and the patch is a bright disc; at midwinter it is 63.4° off, the patch is stretched more than three times as long, and the light that lands in any one square metre is a third as strong.
Fig 2The year recorderSame year, same bench21 Jun · 14.8 h of daylight · north 501 W/m² · south 152 W/m²
The northern channel, and the pen riding it: the year's power on level groundThe two channels cross on the equinoxes and never touch againThe dial: hours of daylight, 0 to 24 the long way roundThe month scale under the chart: the year runs left to rightThe same bench, run for a whole year. The two channels in the panel are not drawn by hand: they are cut to the model's own curves, and the pens are pulled along them by one handwheel, so every mark on the chart is a number the physics has already said. The bright channel is 40° N — 501 W/m² at midsummer, 152 W/m² at midwinter. The dim one is 40° S, the same curve half a year out of step: when the north is hot the south is cold, and the two cross on the equinoxes. The Sun's distance cannot do that. It changes by 3% over the year and it moves the wrong way — Earth is nearer the Sun in January. What changes is the tilt: the same beam, spread over more of the ground, for a shorter day.