The project
The Body House
A house at 2,300 m, organised the way a body is organised.
Nepal, high mid-hills. Four occupants, two floors, 113 m² gross internal on a 70.62 m² footprint. Twenty sheets, published in two stages, drawn at 1 : 100, 1 : 25, 1 : 10 and 1 : 5. Every dimension descends from M = 520 or is measured off the settled 12-point outline; nothing on these sheets is drawn freehand.
- 70.62 m²footprint
- 30.93 mperimeter
- 58.18 m²ground
- 54.84 m²upper
- M = 520modulus
What the drawings are arguing
A 12-point curved outline, 30.93 m of perimeter enclosing 70.62 m². At equal floor area a compact curved plan has roughly 10% less perimeter than the rectangle, which makes the curve a thermal and seismic asset rather than a cost — less wall to lose heat through, and no masonry corner junctions of the kind that fail first in an earthquake.
Inside it, a spine wall runs east–west and stops 810 mm short, dying in one masonry pier that carries flue, floor and stair. The gap it leaves is the plan's only decision point. Sill, seat and stove hearth sit on one line at +1.04 — 2M above datum, 520 above the floor — so the seat top is the sill and the south wall reads as one continuous ledge for 8.6 m rather than as three windows with furniture under them. Everything below that line is mass; everything above is glass or plaster.
Above, a 5.72 m light well runs from the ground floor to a lantern, shut by a six-leaf shutter at the landing worked by one lever: closed at dusk in winter, open by day in summer. The stack that would strip heat in January is the same stack that purges the house in June, which is why it is shuttered rather than glazed or halved.
None of that is visible from outside, and it is not supposed to be. The exterior is the residue of the carving — the method says so in its first two lines, and Stage 5 is eight sheets of holding to it.
Stage 3 · Sheets 01–12
The plan, the section, and the sun
Twelve sheets. The plans and sections settle, the shading rule is retested at four dates and rebuilt with latitude in it, the section is compressed at the approach, and the south wall is opened up to show what is living inside its thickness. The last sheet is a note on what would be defended and what is doubted.
The modulus arrives on this sheet and never leaves. M = 520 — the height at which a person sits down on a thing without deciding to, which is the same 520 as the village plinth. Here it is promoted from an observation to the number every other number descends from: plinth 1M, the sill-and-seat line 2M, the upper floor 7M, the terrace 12M, the parapet 14M, stair risers M/3 on M/2 treads.
Sheet 01
Ground floor plan
spine · chest · gut · throat · entry — dimensions in metres, levels above site datum
1 : 100 at A3
FFL +0.52 (1M plinth)
The spine stops short. It runs 7.45 m from the west wall and dies in a 750 × 850 pier that is also the chimney breast — one piece of masonry carrying flue, roof and stair. The 0.81 gap it leaves is the door's other choice: left into the chest, right into the gut.
Seat, sill, one line. The ground sill sits at 2M = 1.04, exactly 0.52 above the floor, so the 1M seat's top surface and the sill are the same line. The seat runs 8.6 m and stops where the wall turns out of the sun.
Curved built-ins, no joinery. Seat, counter and shelves are all 0.35–0.65 bands struck off the inner face at a constant offset — a mason's dimension, not a joiner's, which is the cheap labour here.
Sheet 02
Upper floor plan
the north wall steps in 350 — sleeping sits in the narrow end
1 : 100 at A3
FFL +3.64 (7M)
Sleeping in the narrow end. The 350 step-in happens only north of the waist, so the womb loses 0.94 m of width here and the two flank windows end up close enough to work as a cross-draught in summer — 6.78 m across, one room, no corridor.
The landing is a room. It gets the 2.25 m south window and the lantern, so the one space you pass through four times a day is also the warmest and brightest — 8.9 m² that no one has to book.
Children's rooms are deliberately small. 8.4 m² each, each with one window and one wardrobe band. The surplus went to the landing, on the assumption that children at 2,300 m live in the warm room and sleep in the cold one.
The section is the primary drawing, because it is the only one that shows height, and height is what makes a room. Where it disagrees with the plan, it decides.
Sheet 03
Section A–A · north–south
through the throat, the spine and the lantern — sun at 27.7° N
1 : 100 at A3
27.7° N · winter 38.9° · equinox 62.3° · summer 85.7°
The brows are still slab edges — but the upper one is now 210, not 450. Tested at four dates rather than two, the 450 was shading a third of the upper glass at equinox noon. Both edges are now cut to one rule: projection = 0.51 × the drop from soffit to head. See Sheet 06.
Winter sun lands on the floor, not the wall. Measured off the brow tips at 38.9°: 2.08 m of floor downstairs, 1.57 m upstairs, both beginning 0.65 m inside the glass — which is where the seat and the landing bench are, and why the floor is the mass that stores it.
The throat is 5.72 m of open air — and now it has a lid. Ground floor to lantern glass, with a five-leaf shutter at the landing edge that closes it in winter and opens it in summer. Sheet 08.
Sheet 04
Section B–B · east–west
through the stair, looking north — the entry sequence in true height
1 : 100 at A3
18R @ 173 · going 260 · 2R+G = 606
18 risers at 173, going 260. 260 is M/2, so each flight of nine is exactly 2.08 — 4M, the same number as the lobe head and the half-landing level. The pair fits the 3.20 throat with 1.12 m of landing, which is also the only place on the stair you can put down a basket. 2R + G = 606.
Headroom is 2.06, not 2.00. Tight but legal-feeling, and it only pinches over the bottom three treads of flight B where you are already ducking under the landing.
The stair runs against the spine. Flight B sits 375 mm off the spine wall, so the masonry that carries the floor also carries the stair — one thing doing two jobs, which is the argument for the spine being masonry at all.
Sheet 05
The approach, in five stages
what the body does between the field gate and the fire — 21 m, four changes of direction
1 : 250 at A3
read left to right
1 · Seen
From the field track, 21 m out, you see the east flank and the stone lobe standing off it. The house is side-on: no front door on show, nothing to aim at yet.
2 · Narrowed
The cheek wall — 1M thick, planted on top — squeezes the path to 1.35 m. Two people can't pass. This is where you stop being outside.
3 · Held
Into the lobe: 3.4 m², stone floor, 2.08 to the ceiling, unheated. Boots, loads, wet things. One step up (+0.26) and the door behind you can shut before the next one opens.
4 · Turned
The inner door sits at the waist, on the narrowest part of the plan, and faces 40° off the outer one. You must turn your shoulders to pass — the building's own hinge.
5 · Released
Into the throat: 5.72 m of height, daylight from the lantern, and the choice — left to the fire in the chest, right past the pier to the gut. Nobody directs you.
The sheet the whole project turns on. Both brows are tested at four dates rather than two, because shading that passes at both solstices can still fail through the shoulder seasons — and this one did. What it caught, and what it cost.
Sheet 06
Sun on the glass, four dates
both brows tested at winter solstice, 1 February, equinox and summer solstice — not only at the solstices
1 : 25 at A3
shadow line = soffit − projection × tan (altitude)
| Date, solar noon | Altitude | Ground brow 350 · head + 2.65 | Upper brow 450 · head + 5.51 | Upper brow 210 · adopted |
|---|---|---|---|---|
| 22 Dec | 38.9° | 3.058 · clears 408 mm | 5.577 · clears 67 mm | 5.771 · clears 261 mm |
| 1 Feb | 44.7° | 2.993 · clears 343 mm | 5.494 · cuts 16 mm | 5.732 · clears 222 mm |
| 20 Mar | 62.3° | 2.673 · clears 23 mm | 5.083 · cuts 427 mm — 32 % of the glass | 5.540 · clears 30 mm |
| 21 Jun | 85.7° | no sun on the glass at all | fully shaded | 3.121 · 1.04 below the sill |
The rain answer — a throated drip band
A separate stone band over the upper head only: 120 high, underside + 5.70, projecting 80, throated. Your + 5.66 puts the equinox line at 5.508 — it cuts the head by 2 mm and makes the band, not the brow, the governing element. At + 5.70 it clears by 38, the brow keeps the equinox, and the two change places about a week before it. The band shades nothing on any of the four dates.
| Date | Alt | Brow 210 | Band 80 at 5.70 | Governing line |
|---|---|---|---|---|
| 22 Dec | 38.9° | 5.771 | 5.636 | 5.636 · clears 126 mm (band) |
| 1 Feb | 44.7° | 5.732 | 5.621 | 5.621 · clears 111 mm (band) |
| 20 Mar | 62.3° | 5.540 | 5.548 | 5.540 · clears 30 mm (brow) |
| 21 Jun | 85.7° | below the sill | 4.626 | fully shaded |
The rule carries its latitude
Written properly: projection = drop ÷ tan (90° − φ), where 90° − φ is the equinox noon altitude. At 27.7° N the coefficient is 0.525. Both brows are cut just under it — 0.507 on the ground, 0.488 above — and that shortfall is exactly the 23 and 30 mm of equinox clearance. The 0.51 I gave in Stage 3 was the built average, not the rule.
| Latitude φ | 90° − φ | Coefficient | On 690 / 430 drop |
|---|---|---|---|
| 15° N | 75° | 0.268 | 0.185 / 0.115 |
| 27.7° N — here | 62.3° | 0.525 | 0.362 / 0.226 |
| 40° N | 50° | 0.839 | 0.579 / 0.361 |
| 52° N | 38° | 1.280 | 0.883 / 0.550 |
At 52° N the same rule wants 1.28 × the drop — 880 of oversail on this head, four times ours. The number is not portable; the rule is.
The rule, not the number. Projection = drop ÷ tan (90° − φ). The ground brow already obeyed it — 350 on a 690 drop. The upper did not: 450 on a 430 drop is a far deeper cut, which is why it failed in the shoulder seasons while the ground brow passed by 23 mm.
Where the year turns. The ground brow first touches its head at 63.1° altitude — 24 March; the upper at 64.0° — 26 March. Two days apart, and both give the glass back on the matching September dates. At 450 the upper was losing its head on 28 January, in the coldest month of the year.
What 210 cost, and what it now costs. 240 mm less eaves over the upper glass — answered by the 80 band above. But the answer is a second element on a wall that had none, 5.7 m up on the only elevation anyone sees. That is the live doubt: Sheet 12.
Sheet 07
The approach, compressed in section
head height and plan width at all five stages — the lobe ceiling comes down from 2.60 to 4M
1 : 100 at A3
distance measured along the walked line
| Stage | Where | Head clear | Plan width | Section you occupy |
|---|---|---|---|---|
| 1 · Seen | 21 m out, on the field track | open sky | 1.80 | — |
| 2 · Narrowed | the cheek-wall gap | open sky | 1.35 | — |
| 3 · Held | the lobe | 2.08 (4M, was 2.60) | 1.56 | 3.24 |
| 4 · Turned | the waist door | 2.08 (4M, was 2.13) | 0.91 | 1.89 |
| 5 · Released | the throat | 5.72 | 3.20 | 18.30 |
4M, not 5M. 2.08 in the lobe against 2.82 in the chest is a 740 mm drop — a head-and-shoulders difference you feel going in and coming out. At 2.60 it was 220 mm, which no body registers. The ceiling is the underside of the lobe roof slab; nothing structural moves.
The waist door head comes down with it. 2.08 above the ground floor, + 2.60 to datum, so from the lobe lintel to the throat nothing is higher than 4M above your feet. Just under three metres of walking under one head height, and then it goes.
The release is 2.75× in height and 9.7× in section. 2.08 → 5.72 overhead; 1.89 m² → 18.30 m² of cross-section, in one step through a 420 wall. That step is the whole argument for the throat.
Sheet 08
The landing shutter, worked both ways
the stack that strips heat upstairs in January is the stack that purges the house in June — one device, two seasons
1 : 25 at A3
six leaves · 1.975 × 0.520 × 0.090 · void 3.20 × 1.975
Free area
| Open | 6.32 m² — 3.20 × 1.975 |
| Closed | 0.047 m² — 24 mm slot, 1.975 long |
| Ratio | 134 : 1 |
| Summer purge | 0.75 m³/s ≈ 14 ach at ΔT 6 K |
Chest, January evening, 21:00
−3 °C outside · 4 kW stove · air 20 °C at 1.70 m
| Ankle, 0.10 | Landing | |
|---|---|---|
| Shutter open | 13.5 °C | 23.5 °C |
| Shutter closed | 17.5 °C | 16.5 °C |
| Difference | + 4.0 K | − 7.0 K |
How it is worked. A 620 mm oak lever on the landing balustrade at + 4.54 — 900 above the landing floor — on a steel shaft under the nosing; a push-rod turns all six leaves together — the last one is the stair mouth, and it folds behind you as you come up. 90° of lever is the whole travel, it latches at both ends, and there is no half position. It is on the side you pass going to bed, and it is the only moving part in the house.
Winter: shut at dusk, open in the morning. Closed, the ankles in the chest gain 4 K and the landing loses 7 — the right trade once the children are in bed, the wrong one at eight in the morning. The 24 mm residual slot is deliberate: the throat never goes fully dead, so moisture from the stove and the kitchen still finds the lantern.
Summer: open by day, and the lantern with it. Two top-hung lantern sashes give 1.24 m² of outlet against 1.10 m² of inlet at the north ground windows, over a 5.72 m stack. At a 6 K difference that is 0.75 m³/s — about fourteen air changes an hour. The section that costs heat in January is what makes June bearable without a fan.
Principle 5, at 1 : 10. Eight named pockets in 420 mm of wall, two residual wedges filled solid and labelled as such. A leftover you have decided to fill is a detail; a leftover you have not is a defect.
Sheet 09
The wall lived in — south wall, enlarged plan
every pocket in the 420 has a name; whatever is left over is filled solid
1 : 20 at A3
cut at + 1.60, looking down
| Where | Depth | Level | What it is for | ||
|---|---|---|---|---|---|
| A | Sill-seat | south wall, continuous 8.60 | 520 | +0.52 → +1.04 | the seat top is the sill; the datum the whole wall is built from |
| B | Wood store | south wall, chest end, 1.56 | 380 | +0.60 → +0.98 | open, under the seat, arm’s reach of the stove |
| C | Bedding drawers | south wall, two runs of 1.56 | 380 | +0.60 → +0.98 | four timber drawers running on the seat frame |
| D | Book recess | pier between the first two windows | 220 | +1.30 → +2.44 | five shelves at 260 — the only wall you can read from a chair |
| E | Lamp shelf | five window jambs | 220 | +1.56, 120 high | a stone ledge returned into the reveal; 3M, above a seated head |
| F | Hand niche | either side of the waist door | 240 | +1.04 → +1.56 | on the sill-seat line: keys, gloves, whatever is in your hand |
| G | Deep store | waist, gut side, 0.90 wide | 280 | +0.52 → +2.60 | full height to the 4M head; doors flush with the plaster |
| H | Boot bench | lobe wall, 1.56 | 420 | seat +0.78 | open slot under for boots, over the drained stone floor |
| — | Filled solid | two wedges under 150 deep | — | — | a 90 at the throat and a 130 at the waist: too shallow to use, so filled |
The rule that governs all of it. The wall is 150 stone / 120 wood-fibre / 150 earth block. The insulation line is continuous and never cut. Pockets 220 and shallower come out of the inner leaf; anything deeper — seat, wood store, boot bench, deep store — is a local swelling of the inner leaf into the room, not a hole toward the weather.
1.96 m³ and 4.5 m² out of the thickness. That is the whole storage of the ground floor and the whole seating of the south side, and none of it is furniture. There is nothing freestanding against the south wall in this house.
Two wedges are filled. Where the curve leaves less than 150 of usable depth the poché is simply solid — drawn dark, labelled, and named. A leftover you have decided to fill is a detail; a leftover you have not is a defect.
Sheet 10
The waist, and the wall in section
the narrowest 1.9 m of the plan, and the south wall cut from the plinth to the parapet
1 : 20 at A3
section on the chest; waist plan cut at + 1.60
The waist is where the wall is thickest in effect. 420 of masonry on a tight double curve leaves poché that is 420 in one place and, measured across the turn, over 600 in another. That surplus is exactly where the deep store and the two hand niches sit — the plan’s narrowest point is also its densest storage.
The boot bench belongs to the lobe, not the house. 420 deep, seat at +0.78, open below, over a drained stone floor. It is the only pocket on the cold side of the insulation line, which is why it is stone through rather than a swelling of the inner leaf.
The section says the same thing vertically. Everything below +1.04 is mass with something kept in it; everything above is glass, plaster or shelf. The seat, the wood store, the lamp shelf and the book recess all read off that one line, on both floors.
The first drawing of this house made from a position a body can actually occupy. Everything before it was an orbit — which was hiding things.
Sheet 11
Eye level, 1.60 m
the two views a body actually has of this house — no orbit, no three-quarter, nothing from the air
true perspective from the walked line
eye 1.60 · figure 1.72
Stage 1 — 21 m out, from the field track
Station point 21.0 m east of the lobe door, 34° cone. The house is side-on and low: the eye sits well below the sill-seat line, the two brows read as the only horizontals, and the stone lobe stands off the flank. The cheek wall — 1.60 high, exactly eye level — cuts the plinth. No door is on show. The figure is at the squeeze, 19.5 m away.
Stage 2 — in the squeeze, 1.35 m between
Station point 0.65 m short of the gap, 72° cone — about what you take in at that distance. The cheek walls close both edges at eye height, the lobe fills the middle with its door already swung off to the right, and the house above runs out of the top of the frame. From here you cannot see the top of your own house. That is what the squeeze is for.
Both are projected, not drawn. Same outline, same levels, same openings as the plans — pinhole projection at 1.60 m with back faces culled. If the plan moves, these move.
What the orbit was hiding. From 21 m the house reads as two horizontal bands and a stone lump, not as a curved plan. The womb shape is invisible from every position a person will ever stand in — it is felt in the body between stages 2 and 5, and seen almost nowhere.
The brows are the elevation. With the upper cut to 210 the two shadow bands are no longer equal, and the house gains a slight top-lightness it did not have at 450. That is a consequence of a thermal decision, and I think it is an improvement.
Sheet 12
What I'd defend, what I doubt
Decision 01
The spine stops 810 mm short
A full-length spine would give two clean halves and a shorter floor span. Stopping it makes one masonry pier that carries flue, floor and stair, and turns the gap into the plan's only decision point. The cost is a 4.11 m span on the gut side, which is at the limit for timber joists at this depth.
Decision 02
Sill, seat and stove hearth on one line at +1.04
2M above datum, 520 above the floor. It means the seat top is the sill, so the south wall reads as one continuous ledge for 8.6 m rather than as three windows with furniture under them. Everything below that line is mass; everything above is glass or plaster.
Decision 03
Both brows are slab edges
350 at the floor, 210 at the roof, tapering to nothing where the wall turns out of the sun. Both are cut to one rule — projection = drop ÷ tan (90° − φ), which is 0.525 at 27.7° N — so both surrender the head within two days of each other in March and give it back together in September. Rain is handled separately, by an 80 mm throated band over the upper head.
Decision 04
The throat is shuttered, not glazed or halved
Six leaves at 1M across the 3.20 × 1.975 void, one lever on the landing, 6.32 m² open against 0.047 closed. Winter: shut at dusk — +4 K at the ankles in the chest, −7 on the landing. Summer: open by day with the lantern sashes, 0.75 m³/s. The section survives intact and gains a season.
Least sure
The 210 brow now leans on an 80 mm stone band
The shading is settled — one rule, both brows, four dates, and the throat is shuttered rather than argued about. What replaced that doubt is water. 210 of oversail will not keep driving rain off the upper heads at 2,300 m, so the head now carries a separate throated band: 120 high, underside +5.70, projecting 80. It sheds, it costs no sun on any of the four dates, and it hands the shading back to the brow a week before the equinox. But it is a second element on a wall that had none — bedded, throated, and 5.7 m up on the only elevation anyone ever sees. Built badly it is a rain stain the whole house is judged by. I would build a 1.5 m mock-up of that head, in the real stone, before ordering any of it.
Two conflicts, unresolved
The 810 gap is also the kitchen route. Everything carried from the stove to the table passes a 810 mm slot beside a hot chimney pier. It is either the best-considered threshold in the plan or a burn waiting to happen. Widening it to 1.10 costs the pier its structural depth.
The lobe is 3.4 m² and unheated. Four people, winter boots, loads off the field. It works for two and jams for four. Making it bigger makes it a room, and then it wants heating, and then it stops being the pause that the whole approach depends on.
Corrections carried in this set. The upper brow is 210, not 450 — at 450 it shaded 32 % of the upper glass at equinox noon while clearing winter by only 67 mm. Winter penetration, remeasured off the brow tips at 38.9°, is 2.08 m downstairs and 1.57 m upstairs, not the 2.38 / 2.06 I gave before. The lobe ceiling is 2.08, not 2.60, and the waist door head comes down with it. Latitude is stated as 27.7° N throughout, which is where the 62.3° equinox altitude comes from, and the brow rule is written as projection = drop ÷ tan (90° − φ) so the number carries it. The stair going is 260, not 255 — M/2, and each flight of nine is then exactly 2.08. The waist pockets on Sheet 10 have moved clear of the door swing, which they overlapped in the first issue of this sheet.
Stage 5 · Sheets 13–20
The exterior and the ground
Eight sheets, and not one of them moves a window. The form is settled by the interior and does not move; what is designed here is how the body is dressed — surface, junction, edge — and the ground it stands on.
Sheet 13
Material junctions
every place two materials meet, and the job each junction does
1 : 5 at A3
wall 420 = 20 render / 130 stone / 120 wood fibre / 140 earth block / 10 plaster
J1 · plinth to render, + 0.52
J2 · stone lobe to render — the parting joint
J3 · render to brow soffit
J4 · render to parapet coping
J5 · render to reveal and sill
J6 · where the render stops at the ground
One rule covers five of the six. Every horizontal surface on this house throws water away from the face below it: the plinth weathering, the sill oversail, the drip band, the brow throat, the coping throat. Where a horizontal cannot throw, the render is not there.
The render never meets the ground. It starts at + 0.52 and the plinth takes the whole ground-contact zone — 520 of stone above the apron, 170 clear of the 350 splash line, with a 300 × 200 washed-stone margin at its foot to break the splash before it arrives.
The lobe is not bonded to the render, it is bonded behind it. A 40 open joint, hemp-lime packed and drained at the foot, keeps the two masses reading as two. It is the only deliberate gap on the outside of the building.
Sheet 14
The render
90 % of what anyone sees — three coats of lime, one material, two tools
NHL 3.5 · 8 scratch / 8 float / 4 finish = 20
five coats of limewash, 3 % yellow ochre
Colour
Limewash, five thin coats, pigmented with local yellow ochre at 3 % of the lime by weight — a warm pale earth, not white. At 2,300 m the December light is thin and very blue and a neutral white render goes grey under it; the ochre puts back roughly what the sky takes out, so the wall still reads warm at 38.9°. Under the high June sun the same wall reads almost white. One pigment, two seasons.
finish coat · after ten years · in shadow at noon
Twenty years
Lime carbonates slowly and closes its own hairline cracks, which is why it is here and cement is not. It will darken under any horizontal that does not throw water — and every horizontal on this house throws. What it will not survive is the north-west quadrant: no sun, prevailing wet, and it will green. Accept it as a soft grey-green wash rather than fight it. Re-limewash south and west at about year twelve; north can go the full twenty.
One coat, two tools. The change at + 1.04 is not a joint and not a different material — it is the same continuous three-coat lime, floated open below and sponged closed above. Nothing to fail at the line, and the seat above it casts the shadow that makes the change legible.
Why coarse below. The bottom 520 of every wall in a working house takes boots, baskets, firewood and children. A 4 mm open float takes a scar and hides a patch; a fine closed finish shows every one of them for twenty years.
Limewash, not paint. Five thin coats, re-applied by anyone with a brush and a bucket, no scaffolding tender and no specialist. That is the maintenance argument, and it is the only reason a rendered house makes sense at this altitude.
The apron is a chautari, and nobody set out to draw one. 8.60 × 3.64 m of riven flags with a seat wall whose top sits 520 above them — exactly as the sill-seat sits 520 above the floor inside — and one deciduous cherry. A platform and a tree, arrived at six versions and one site away from the thing this site is named after.
Sheet 15
The south apron, as a place
7M of paving becomes a chautari: a platform and a tree
1 : 100 at A3
apron 8.60 × 3.64 = 31.3 m² at + 0.26
Deciduous is not a preference, it is a requirement. In leaf at winter noon, a 6.00 m crown 2.60 m out would put its shadow on the south wall up to + 3.91 — the whole ground floor and half the upper. Every number in Sheet 06 assumes bare branches from November to March. Prunus cerasoides is bare exactly then, and flowers in October when nothing else does.
The tree is pruned, not planted and forgotten. Single stem, crown lifted to 2.60 so you walk under it, held at 6.00 m. It stands on the pier between the first and second south windows, so it never sits in front of a window head, and it is set in a 1.04 × 1.04 unpaved square — 2M, one flag omitted.
The edge that was missing. A 420 wall on the outer edge, top at + 0.78 — 520 above the apron, exactly the height the sill-seat sits above the floor inside. The same seat, one level down. It runs 5.48 and stops 1.56 short at each end so you can walk off the corners.
The apron is the lobe floor. Both at + 0.26. Two risers of 130 take you up to + 0.52 at the south door, and the flag surface is the same stone on both sides of it — so the apron reads as inside the house that happens to have no roof.
It drains through, not to. Riven flags 60 thick on 100 of compacted stone, butt-jointed with a grit fill, falling 1:60 away. No gully, no channel, nothing to block with leaves in October when the cherry drops.
Sheet 16
The ground, designed
the cut, the wind wall, the water, the wood, and the five surfaces of the path
1 : 100 at A3
bench cut 1.60 north, filled 0.90 south
| Stage | Where | Surface | What it does |
|---|---|---|---|
| 1 · Seen | the field track | rammed earth, no edge | whatever the field is |
| 2 · Narrowed | the cheek-wall gap | three riven flags, 900 × 600, laid loose | the ground firms before the wall does |
| 3 · Held | the lobe | flags 60, close-jointed, fall 1:60 out | slotted stone drain at the door |
| 4 · Turned | the waist door | the same flags, unbroken through the threshold | the level changes, the material does not |
| 5 · Released | the throat | the same stone, 1.04 runner of larger flags on the walked line | polished by use, not by tooling |
The stone gets more worked, never different. Rammed earth, then loose flags, then laid flags, then close-jointed flags, then a polished runner. One material from the field gate to the fire, changing only in how much work has been done to it — which is also the measure of how far inside you are.
Roof water is a store, not an attenuation. 62 m² of roof; 40 mm of monsoon in an hour is 2,480 litres, so a 2,000-litre cistern fills and overflows inside the hour. It is there for the dry months. The overflow goes to a soakaway in the field, not to the drainage run.
The firewood is not against the house. It stands against the retaining wall, 200 clear of it, on 150 of stone, under slate, open at both ends. Four cubic metres is about one winter. Nothing that harbours damp or insects touches a wall that has insulation in it.
Sheet 17
Night
sixteen hours of it in December — four fixtures, none of them pointing up
lantern glass 1.55 × 0.61 at + 6.55
2700 K at the threshold, 3000 K in the lantern
Threshold — one fitting. A lamp set in the wall thickness at the lobe door, in the outdoor twin of the hand niche, at + 1.56 (3M). 300 lm at 2700 K, shielded so nothing above 1.60 is lit. From the field you see lit stone and never the source.
Apron — no fitting at all. It is lit by what spills through the south glass, and the sill-seat line at + 1.04 puts that band of light at exactly the height that grazes the flags. A cast hook in the seat wall every 2M takes a carried lantern when one is wanted.
Stair — one low light. Recessed 220 into the spine at the half-landing, + 2.08, 80 lm, throwing across the treads from the side. Nobody climbs toward a bulb.
Landing — the one that shows. The lantern is lit only by the landing light, and the shutter closed at dusk stops that light falling down the throat and washing out the chest. So the house is at its most visible from the valley at exactly the moment it is most closed.
The stove is the second light. In December it is on for sixteen hours and it is the reason the chest reads warm at 2700 K when everything outside is at 6000. No fitting can do that, and none should try.
Four fixtures, 680 lumens. Threshold 300, stair 80, landing 200, lantern shared. That is the whole external and circulation lighting of the house, and it is deliberate: at 2,300 m the thing worth seeing at night is outside the window.
Nothing points up. No uplight on the wall, no light in the soffit of either brow, nothing on the parapet. The rendered wall at night is dark, and the only lit thing above 6.24 is the lantern glass itself.
The lantern is not a beacon. 40 lumens leaving a 0.95 m² pane at 800 m is a lit window, not a signal. It reads as somebody being awake, which is the correct amount of information for a house to give a valley.
The exterior is a result. Nothing on these two elevations was moved to balance them, and every horizontal on the south face is a line that throws water.
Sheet 18
South elevation
a record of what the carving produced — nothing here was moved to balance it
1 : 100 at A3
orthographic, from the same outline as the plans
Three windows, not evenly spaced. 1.48 / 2.25 / 1.48 on whole arc widths, sitting where the chest, the throat and the gut want them. The gaps between them are the piers that carry the book recess and the lamp shelves. The elevation is a consequence, and reading it tells you the plan.
Four horizontals and nothing else. Plinth top, sill-seat, brow, drip band, coping. Everything that appears to be composition on this face is a line that throws water.
The upper windows are smaller and higher than they "should" be. That is the 430 drop from soffit to head doing its work, and it is why the upper brow is 210. If the elevation looks top-light, that is the shading rule showing, not an error.
Deliberately unbalanced: the lobe door faces the track, not the house.
Sheet 19
East elevation
the face you actually arrive at — stone, and a gap
1 : 100 at A3
cheek wall shown in front, cut at 1.60
This face has almost no glass, and that is the plan speaking. The east is the entry side and the cold morning side; the gut window and one bedroom window are all it gets. The lobe does the work of a front elevation without being one.
The 40 joint is the whole composition. Two masses, two materials, one gap between them, drained at the foot. There is no moulding, no band and no change of plane on this face other than the ones the water requires.
It looks unbalanced, and it is. The lobe sits off-centre because the door faces the track, not the house. Moving it would balance the elevation and break the approach. Sheet 11 shows what you actually see from here, which is not this drawing.
Sheet 20
Three decisions, and the one I would not bet on
Decision 05
The plinth is the only thing that touches the ground
Render begins at + 0.52 and never goes lower. The plinth stands 150 proud of the render face, 520 high above the apron, top weathered 1:6 with a 20-back throat, on a 300 × 200 washed-stone margin. Splash at this altitude reaches 350; the render starts 170 above it. No render, anywhere, is within reach of a boot or a downpour.
Decision 06
One render, two tools, changing at + 1.04
Three-coat NHL 3.5 — 8 / 8 / 4 = 20 — continuous top to bottom. Below the sill-seat line it is floated open on 4 mm sharp sand; above it is closed with a sponge on 2 mm. No joint, no bead, nothing to fail. Five coats of limewash at 3 % yellow ochre, which is what keeps the wall warm under a December sky at 38.9°.
Decision 07
The apron becomes a chautari
8.60 × 3.64 (7M) of riven flags at + 0.26 — the lobe's level — edged by a 420 wall whose top at + 0.78 is a seat, 520 above the apron, exactly as the sill-seat is 520 above the floor inside. And one deciduous cherry, 2.60 out, crown held at 6.00, which shades 17 of the 31 m² at June noon and is bare for every day that matters in Sheet 06.
Least sure
The tree, because it is the only element I cannot draw
Every other decision on these eight sheets is a detail I can set out at 1:5 and check. The cherry is a thirty-year bet on somebody pruning. It must stay deciduous, it must stay under 6.00, and it stands 2.60 from a rendered wall on a cut bench with its roots beside a plinth. If it is allowed to reach 10 m it shades the whole south wall to + 3.91 at winter noon and quietly cancels Sheet 06 — the sun geometry that four rounds of this project have been built on. The alternative is a demountable timber and canvas frame over the apron, taken down in October, which does the June job with no thirty-year risk and none of the reason to sit there. I have drawn the tree, and I would write the pruning regime into the drawings rather than hope.
Two things I did not resolve
The north-west quadrant will green. No sun, prevailing wet, lime render. I have accepted it and specified re-limewashing south and west at year twelve and north at twenty. The honest alternative is stone on that quadrant and render on the other three, which would be truthful about exposure and would also make the house look like two buildings.
The east elevation is unbalanced. The lobe sits off-centre because its door faces the track. That is information about the approach, not a fault in the elevation, and I have left it — but it is the one face where somebody will ask to move something, and the answer has to be no every time.
Carried forward unchanged: the outline, LEVELS, M = 520, both brows at 350 and 210, the drip band at + 5.70 projecting 80, the sill-seat-hearth line at + 1.04, the spine and the 810 gap, the throat and its six-leaf shutter, and every shadow line verified in Stage 4.
Where this and the village disagree
The village house is a rectangle, 4.6 × 6.2 m, with its gable to the path, because at that size a building reads as a house from its gable and as a shed from its long side. This one is a curved plan under a flat, inhabited roof. Those are different answers.
They are also different questions: a unit that has to be repeatable forty times on one hillside, against one family's house at 2,300 m. Both were arrived at by the same method and neither has been restyled to agree with the other.
Build it with us