# Slow Bloom — concept sheet

**Status: APPROVED 2026-11-01 — Mike's five decisions are applied in this revision. NOTHING IS
ORDERED AND NOTHING IS CUT.** The bill of materials is updated to the parts he chose; the
structural check comes before any metal is cut and it is a licensed engineer's job — the package
that check needs is `docs/slow-bloom-structural-brief.md`.

Direction chosen by Mike on 2026-11-01 from `docs/concepts.md`. Issues: NDY-2 (concept sheet),
NDY-3 (decisions applied). Prices read 2026-11-01. Machine-readable bill of materials:
`docs/slow-bloom-bom.json`, checked by `scripts/slow_bloom_bom.py`. Structural inputs:
`output/slow-bloom-load-inputs.json`, checked by `scripts/slow_bloom_loads.py`
(see *Verification* at the end).

---

## 1. What it is

A single dark stem rises out of the lawn and carries one long arm. On the arm, five shallow
stainless bowls — each cut in half and re-set on opposite sides of its own little axle — turn
at their own speeds in the wind, so the arm looks like a flowering branch stopped mid-swing.
The arm swings to face the wind, leans a few degrees into a gust and eases back, and the five
blooms spin. Nothing hangs. Nothing is powered. Nothing is asked to do anything but turn.

The whole piece is one stalk, one arm, five blooms, three motions. That restraint is the design.

---

## 2. Form

### 2.1 The scale, in numbers

| | |
|---|---|
| Overall height | about 11 ft 6 in |
| Overall reach from the mast | about 12 ft 4 in |
| Mast | 4 in Schedule 40 pipe, 7 ft 6 in clear above grade |
| Boom | 12 ft long, 2 ft 2-1/2 in OD tube, balanced on its pivot 4 ft from the butt |
| Blooms | 5, each 13-1/4 in across and 4 in deep |
| Bloom centres along the boom | 6 ft 0 in, 7 ft 5 in, 8 ft 10 in, 10 ft 2 in, 11 ft 7 in from the butt |
| Bloom centres above the boom | 42, 36, 30, 24, 18 in, inner bloom highest, tip bloom lowest |
| Weight, moving parts | about 155 lb, of which 50 lb is the counterweight — itemised in the structural brief |
| Weight, fixed | about 128 lb (mast 81 lb, base plate and gussets 39 lb, fixed head plate and drip hat 7 lb) |
| Footprint | one 12 in diameter pier, 2 in proud of grade |
| Design wind | 115 mph, 3-second gust, Exposure C (Grand County) — no structural check has been done |

The bloom centres wander rather than line up; no two are at the same height or the same
distance out. That is deliberate — it is what makes the cluster read as a flower rather than a
row of cups on a fence.

### 2.2 The stem

The mast is a plain 4 in pipe, painted the same dark colour as the arm and left unadorned. It
should read as a stem seen from the side, not as a post. It is 7 ft 6 in high, which puts the
whole mechanism — the ring, the yoke, the counterweight — above eye level: you look at the
piece from underneath, at the blooms, and the machinery stays out of the picture.

The foot is one 12 in diameter concrete pier, **36 in below finished grade** (the frost depth in
Grand County's adopted code — see the reference sheet) **plus the 2 in that stands proud**, so 38 in
of form and pour in total; the concrete is brushed and left plain. A 12 in square, 3/4 in thick
steel base plate is welded to the bottom of the mast with four triangular gussets and held to the
pier with four 3/4 in screw anchors. From three feet away the foot is a plate on a disc of concrete,
and that is all it should be.

*Correction made in this revision:* the earlier draft cut the 4 ft form tube down to 3 ft and then
struck the top 2 in proud, which leaves only 34 in of pier in the ground — 2 in short of the
36 in the adopted code requires. The tube is now cut to 38 in. Whether 36 in of 12 in diameter
pier is *enough* is a different question and it belongs to a licensed engineer; see
`docs/slow-bloom-structural-brief.md`.

### 2.3 The arm

The boom is 12 ft of 2-1/2 in tube. It is **straight** — the taper in the silhouette comes from
what is mounted on it, not from the tube: the blooms crowd the outboard end, their stalks get
shorter toward the tip, and the 50 lb counterweight hangs on the inboard 4 ft. The eye reads a
tapering branch; the metal is a straight tube, which is cheaper, stiffer and honest.

The pivot is 4 ft from the butt, so the arm is 4 ft one way and 8 ft the other. The butt carries
a short 1 in stub welded in line with the tube, with two 25 lb cast-iron plates slid onto it as
the counterweight. Sliding them sets the balance; a shaft collar locks them. The intent is a boom
that sits very slightly bloom-heavy, so the springs (below) are always just preloaded.

### 2.4 The blooms

Each bloom starts as one 13-1/4 in spun stainless bowl, 4 in deep — the **thin 0.4 mm** 8 qt bowl,
confirmed by Mike on 2026-11-01 as the chosen gauge: motion over toughness. It will dent. At $17.29
each that is the trade he accepted and it is not re-argued here. It is **cut in half on its
diameter**, and the two halves are re-mounted on opposite sides of the bloom's own vertical
axle: convex sides out, cut edges facing each other, each half pushed 1-1/2 in off the axis in
opposite directions, with about 1-1/4 in of clear space between them.

That offset is the whole mechanism. Two offset shallow lobes on a vertical axle are a drag
rotor — a Savonius-type machine — which starts turning from rest in a light breeze instead of
needing a threshold of wind like a propeller does. It is a slow, high-torque, inefficient
machine by nature, and slow is what this piece wants. The blade research on the reference sheet
puts a Savonius drive's power coefficient at 0.15–0.25 and its tip-speed ratio near 1 — a garden
Savonius is a slow-turning, high-torque device, which is exactly the motion called for.

The two halves are the petals: from the ground you see two bright curved crescents with a clear
eye through the middle, turning. The bowl's inside faces are painted a flat contrasting colour
to the outsides, so the bloom flickers dark-light as it turns — the same dodge the Wind Curtain
direction used, reduced to two faces instead of hundreds.

### 2.5 Surface and colour — decided

**Colours (Mike's decision 1, 2026-11-01): pairing 1.**
**Charcoal-green stem and arm, marigold blooms, clay-red bowl interiors.** The warm one: legible
from the street, and the red interiors give the strongest dark-light flicker as each bloom turns —
which is the whole reason the bowls are painted inside and out in two different colours. The
other two pairings considered at NDY-2 were black/bone/slate-blue (the quiet one) and
oxide-red/celadon/bone (the most plant-like); both are recorded and both are rejected.

The bowls themselves are bare stainless outside where they are cut; only the armature and the
bloom exteriors take the body colour, and the bowl interiors take the counter-colour.

**Finish (Mike's decision 3, 2026-11-01): one single-can direct-to-metal enamel, $53.98.**
Not the two-part marine LPU over a two-part epoxy primer. Being direct-to-metal there is no primer
line, and the two marine lines are gone from the bill — **$489.98 removed, $53.98 put back,
$436.00 off the total**. The consequences are stated rather than glossed:

- **A shorter service life than the boat system.** This is a single-component oil-based enamel.
  The two-part LPU it replaces was chosen because it is what survives sun, freeze-thaw and grit on
  a wet dock. The enamel will chalk and it will need recoating, and the honest expectation is a
  garden piece rather than a dock piece. Mike accepted that trade knowingly, and this sheet does
  not re-argue it.
- **No primer, and none needed.** Direct-to-metal means the prepared steel is the substrate.
  Degrease and abrade it properly and that is the whole coating system.

**One open item this creates — coating colour.** Decision 1 asks for three colours; decision 3
buys a product that cannot be tinted. Rust-Oleum's own listing for the Professional High
Performance Protective Enamel says it "cannot be tinted", and offers its own workaround — mix
colours from the same product line and sheen to make custom colours — while the flat sheen in that
line is stocked only in black and white. So charcoal-green, marigold and clay-red are not three
cans off a shelf. Three ways out, in order of preference:

1. **Mix from stock gallons** of the same product and sheen, using Rust-Oleum's own method. Most
   faithful to the chosen colours; costs more than one can, and the exact mix has to be recorded
   so it can be matched on the recoat.
2. **Accept the nearest stock colours** in the line (a dark green in gloss, safety yellow, a red).
   Cheapest and simplest; the sheens then differ between the armature and the blooms unless a
   single sheen is chosen throughout.
3. **Move the colour coats to a tintable single-component direct-to-metal enamel** at similar
   cost, keeping decision 3's system (single-component, DTM, one can per colour) and getting the
   colours exactly.

The bill carries the $53.98 flat-black gallon as the price basis for the decision, which is what
Mike decided against. The colour coats get priced once this is settled — it is the only place where
two of his five decisions pull against each other, so it is named here rather than quietly resolved.

The bowls themselves are bare stainless and need no paint at all; only the armature and the bloom
exteriors are coated.

### 2.6 How it reads from thirty feet

From the street the piece is a dark, still calligraphic mark against the house, with five bright
loops hung off one long arm. What moves at that distance is the **whole branch** — the slow
grand yaw as it swings to face a wind change, and the lean. You read direction and strength of
the wind off it before you read anything else: the arm points downwind, and how hard it leans
tells you how hard it is blowing. The blooms are a shimmer at that range, not a subject.

### 2.7 How it reads from the patio

Close up the blooms are the whole thing. Each is a shallow bright bowl split down the middle,
spinning on its own little stainless axle, and because each rotor's drag and inertia differ
slightly, no two turn at the same rate — so the cluster never syncs and never stops being
interesting. You can see the gap between the halves, the welded hub, the two screws that hold
each rotor on. Sitting six feet away the piece is five slow wheels of colour and one long
straight bar, and it is legible as a machine and as a flower at the same time, which is the
point.

---

## 3. Motion — three speeds at once

| Motion | What drives it | Speed in a 5 mph breeze | Reads best from |
|---|---|---|---|
| **Yaw** — the whole branch swings to face the wind | wind pressure on the blooms, which all sit on one side of the mast | a full turn in roughly 15–60 s | the street |
| **Lean** — the arm tips into the gust and eases back | the change in wind moment on the blooms, taken against four urethane pads | 3–8° of movement, period roughly 3–6 s, always returning to the same attitude | the street, and the patio |
| **Spin** — each bloom turns on its own vertical axle | local flow at each rotor (two offset lobes, drag rotor) | 5–30 rpm, each bloom different | the patio |

**The lean is the piece's signature and its one non-obvious mechanism.** The boom does **not**
rock free on a balance point. It leans against four urethane pads (part C3) that are set as
combined stops and springs, so the arm can move only 3–8° and always eases back to the same
attitude when the gust passes. It hunches and it recovers. That is a very different motion from
a swinging arm, and it is what keeps the piece from being a Calder with a post bolted on — see
Risk 3.

In a dead calm nothing moves at all and the piece holds one fixed silhouette: a flower frozen
mid-bloom. In a hard wind the arm leans hard against the pads and the blooms blur. There is no
stop on the spin (a rotor never needs one) and no stop on the yaw (full rotation).

---

## 4. Materials shortlist

Every line below has a real part number and a source link, and every line carries a price — except
one, which cannot, and says so. Notice what is
conspicuous: there are no structural quantities in the list, no castings, no machined parts and
no custom spinning — the armature is tube and plate welded together, and every rotating joint is
an off-the-shelf bearing. Nothing on this list requires a supplier to make anything for us.

**Totals** (`scripts/slow_bloom_bom.py`, run 2026-11-01 on the updated bill): **36 lines, all 36
primary, $2,396.48** before tax and shipping. That is **$436.00 less than the $2,832.48** the bill
stood at before Mike's decisions, and the saving is almost exactly what decision 3 predicted.
30 of the 36 prices were read directly off the vendor's own page; 5 are marked `listing`; 1 is
marked `quote` — the sealed main ring, which is the one line in the bill that still has no real
price against it. There are no alternate lines left: the finish that used to be the alternate is
now the finish.

**The total is given two ways, because the main ring has no published price.** The $2,396.48 above
carries the sealed ring at a **floor allowance** of $398.83 — the published price of the same-size
standard ring, which the sealed one cannot undercut. If the sealed ring instead costs what the only
published sealed/standard pair in the igus range implies (the 80 mm sizes, where the sealed ring is
2.63× the standard one), the ring lands near **$1,050** and the bill becomes **$3,047.88**. So the
honest headline is: the finish decision saves $436, and the ring decision may give back more than
that. Until igus puts a number in writing, the figure for this ring is an allowance and not a price.

| ID | Item | Part number | Price | Qty | Vendor |
|---|---|---|---|---|---|
| A1 | Concrete mix, 80 lb bag (Quikrete 1101) | 110180 | $6.47 | 5 | The Home Depot |
| A2 | Form tube 12 in x 4 ft (Sonotube) | Menards Sku 1895159 | $17.38 | 1 | Menards |
| A3 | Rebar #4 x 4 ft, A615 Gr 40 | RB12 | $7.80 | 4 | Metals Depot |
| A4 | Steel pipe 4 in Sch 40 x 8 ft (mast) | T54 | $361.00 | 1 | Metals Depot |
| A5 | Steel plate A36 12 x 12 x 3/4 in (base plate) | 9614 | $232.21 | 1 | Online Metals |
| A6 | Steel angle 2 x 2 x 1/4 in x 10 ft (gussets) | A12214 | $64.50 | 1 | Metals Depot |
| A7 | Titen HD screw anchor 3/4 x 8-1/2 in, box of 5 | THD75812H | $59.09 | 1 | Concrete Fasteners |
| B1 | **iglide PRT-01-100-DD slewing ring, fully sealed** (the main thrust bearing) | PRT-01-100-DD | $398.83 ⚠ allowance | 1 | igus |
| B2 | Steel plate A36 12 x 12 x 1/2 in (head plates) | 9612 | $78.55 | 1 | Online Metals |
| B3 | 1 qt stainless mixing bowl → the drip hat over the ring | Staples 453546 / model 78701 | $8.99 | 1 | Staples |
| C1 | **UCFL205-16 two-bolt flange bearing unit**, 1 in bore, sealed | UCFL205-16 | $24.89 | 2 | Zoro |
| C2 | 1 in 1018 cold finish round bar, 12 in (nod pin + cwt stub) | 4797 | $11.62 | 2 | Online Metals |
| C3 | 1 x 2 x 1 in urethane bumper, 80A (lean springs/stops) | REC003 | $7.27 | 4 | Acrotech |
| D1 | Steel tube 2-1/2 in OD x 0.120 x 20 ft (the boom) | T2W212120 | $150.60 | 1 | Metals Depot |
| D2 | Steel tube 1-1/2 in OD x 0.120 x 12 ft (bloom stalks) | 866TF9 | $140.00 | 1 | Grainger |
| D3 | Steel tube 1-1/4 in OD x 3/16 DOM x 6 ft (stalk collars) | T2114316 | $72.30 | 1 | Metals Depot |
| D4 | CAP 25 lb cast iron plate, 1 in bore (counterweight) | Walmart item 20470506 | $34.97 | 2 | Walmart |
| D5 | Plastic tube end caps 2-1/2 in, 12 pcs | Lowes 8711039 | $11.99 | 1 | Lowe's |
| D6 | Plastic tube end caps 1-1/4 in, 40 pcs | Lowes 8059932 | $12.99 | 1 | Lowe's |
| E1 | **8 qt stainless mixing bowl, 13-1/4 in x 4 in, 0.4 mm** (the bloom — chosen) | Staples 454113 / model 78705 | $17.29 | 5 | Staples |
| E2 | Stainless tube 316 1 in OD x 0.065 x 0.87 ID (bloom hub) | Online Metals 12960 | $18.40 | 3 | Online Metals |
| E3 | 8 mm x 100 mm hardened chrome shaft, 5-pack (axles) | VXB 8mm-100mm-hardened | $49.93 | 1 | VXB Bearings |
| E4 | **608-2RS sealed bearing 8 x 22 x 7 mm, 10-pack** (spin bearings) | VXB 10-pack 608-2RS | $14.99 | 2 | VXB Bearings |
| E5 | Steel plate A36 1/4 in x 12 x 12 in (hub caps + yoke cheeks) | 9610 | $27.55 | 2 | Online Metals |
| E6 | Stainless tube 316 3/4 in OD x 0.049 x 3 ft (sleeves, spacers) | Online Metals 12952 | $39.44 | 1 | Online Metals |
| E7 | Steel flat bar 1/4 x 2 in x 6 ft (clamps, tabs) | F2142 | $26.18 | 1 | Metals Depot |
| F3 | **Direct-to-metal enamel, gallon** (the finish — no primer line) | Rust-Oleum 7776402 | $53.98 | 1 | The Home Depot |
| F4 | SS 316 socket cap screw 1/4-20 x 1 in | Bolt Depot 16540 | $0.59 | 48 | Bolt Depot |
| F5 | SS 316 socket cap screw 5/16-18 x 1-1/4 in | Bolt Depot 16550 | $1.04 | 24 | Bolt Depot |
| F6 | SS 316 hex nut 1/4-20 | Bolt Depot 15983 | $0.12 | 48 | Bolt Depot |
| F7 | SS 316 hex nut 5/16-18 | Bolt Depot 15984 | $0.16 | 24 | Bolt Depot |
| F8 | SS 316 flat washer 1/4 in | Bolt Depot 16019 | $0.14 | 96 | Bolt Depot |
| F9 | SS 316 flat washer 5/16 in | Bolt Depot 16020 | $0.16 | 48 | Bolt Depot |
| F10 | SS 18-8 metric socket cap screw M4 x 20 mm | Bolt Depot 6398 | $0.11 | 60 | Bolt Depot |
| F11 | 3M 5200 marine adhesive sealant, 10 oz | PN06500HC | $22.48 | 1 | The Home Depot |
| F12 | Boeshield T-9 corrosion protectant, 12 oz | ASIN B0HNT7FM27 | $21.56 | 2 | Amazon |

Full source links, the exact URL for every line, the per-line sourcing notes, and the
`page` / `listing` / `quote` verification flag per line are in `docs/slow-bloom-bom.json`. Category
subtotals: foundation and mast $797.73 · yaw axis $486.37 · lean axis $102.10 · boom and balance
$457.82 · blooms $342.28 · finish and fasteners $210.18.

Line ids **F1 and F2 are deliberately absent** from the table: those were the two-part epoxy primer
and the two-part marine topcoat, deleted on 2026-11-01 by decision 3. Nothing was renumbered, so the
gap is the record of the removal rather than a lost line. The finish section fell from $646.18 to
$210.18 and is now almost all stainless fasteners.

**Two lines were once unpriced; one of them still is. Both are handled explicitly rather than papered over:**

- The **14 in** concrete form tube used in the first draft of this design is a real product but
  **no vendor publishes a price for it online** (Home Depot, Lowe's and Menards all list it
  store-only). The design was therefore changed to a **12 in** pier, which is priced and buyable
  online at Menards ($17.38). That change stands.
- The **fully sealed main ring, PRT-01-100-DD**, is now the *chosen* part (decision 2) and **still
  has no published price** — igus quotes it on request. It remains **the only line in this bill
  without a real price against it**, it is carried as a $398.83 floor allowance rather than a
  quotation, and the total above is therefore stated two ways. Its 25–30 working day lead time is
  accepted and is a real constraint on the build schedule — see §5.3.

---

## 5. Build plan

### 5.1 Cut list

| Stock (BOM id) | Cut | Qty | Length each |
|---|---|---|---|
| 4 in Sch 40 pipe (A4) | mast | 1 | 90 in |
| 12 in x 4 ft form tube (A2) | pier form (cut down) | 1 | 38 in |
| 2-1/2 in OD tube (D1) | boom | 1 | 144 in |
| 2-1/2 in OD tube (D1) | fulcrum cross-tubes / fixtures | 4 | 6 in |
| 1-1/2 in OD tube (D2) | bloom stalks, inner to outer | 5 | 40, 34, 28, 22, 16 in |
| 1-1/4 in OD DOM (D3) | stalk collars | 5 | 3 in |
| 1 in OD 316 tube (E2) | bloom hubs | 5 | 6 in |
| 8 mm hardened shaft (E3) | bloom axles | 5 | 100 mm |
| 1 in 1018 bar (C2) | nod pivot pin | 1 | 7 in |
| 1 in 1018 bar (C2) | counterweight stub | 1 | 5 in |
| 1/4 in A36 plate (E5) | hub caps, 3-1/4 in dia discs | 10 | — |
| 1/4 in A36 plate (E5) | nod yoke cheeks | 2 | 8 x 6 in |
| 1/4 in A36 plate (E5) | brackets and gussets | 4 | as needed |
| 1/2 in A36 plate (B2) | fixed head adapter plate | 1 | 7 x 7 in |
| 1/2 in A36 plate (B2) | rotating head plate | 1 | 7 x 7 in |
| 3/4 in A36 plate (A5) | base plate | 1 | 12 x 12 in |
| 2 x 2 x 1/4 angle (A6) | base gussets | 4 | 8 in |
| 1/4 x 2 flat (E7) | bowl-to-hub clamps | 5 | 6 in |
| 1/4 x 2 flat (E7) | about 5 in, stub collar blank | 1 | 5 in |
| Rebar #4 (A3) | pier verticals | 4 | 48 in |

**Critical fits and dimensions** — the numbers that have to be right before anything is welded:

- Bloom hub bore: the 1 in OD stainless tube has a 0.87 in (22.1 mm) bore — a light press on the
  22 mm bearing OD. Press dry; if a bore comes up loose, Loctite 638 retaining compound, not a
  bigger hammer.
- Bloom axle: use **only** the ground 8 mm shaft (E3). Ordinary 8 mm bar is undersized and the
  bearing inner ring will creep and wear the shaft.
- Nod yoke cheek spacing: set the cheeks so the boom hub clears them with 1/8 in a side, and fix
  the four bumpers (C3) so they both stop the lean and sit in light compression at rest. Total
  travel is set here: adjust the shims under the bumpers to land the lean at 3–8°.
- Counterweight stub: 5 in of 1 in bar welded into the butt **coaxial** with the boom, spanning
  the 4 ft counterweight arm clean to the end.
- Weep holes: a 3/16 in hole at the lowest point of every closed tube volume (boom ends, stalk
  bottoms, hub underside, the head's plate sandwich). Ends get plastic plugs (D5, D6). A sealed
  tube full of water splits its own welds in the first hard freeze — the plugs stop the water
  getting in, the weep holes stop it staying.
- Drain holes: four 3/8 in holes in the fixed head plate, positioned so no water can stand under
  the slewing ring.
- Bolt pattern for the ring: taken from the igus CAD for PRT-01-100, downloadable from the
  product page. Drill from that, not from a measurement.

### 5.2 Sequence

1. **Set the pier.** Dig 12 in dia x 38 in deep; 4 verticals of #4 rebar wired into a cage;
   Sonotube cut to 38 in; pour, rod, strike the top **2 in above grade so that 36 in of pier stands
   in the ground** — the adopted frost depth — and cure for a week before the piece goes on it.
   **This step waits on the structural check.** The pier is the one part of this build that cannot
   be undone cheaply, and decision 5 puts the engineer's check in front of it.
2. **Mast.** Cut the 4 in pipe to 90 in. Weld the 12 x 12 x 3/4 in base plate to the bottom, then
   the four 2 x 2 x 1/4 in angle gussets. Weld the 1/2 in head adapter plate to the top, squared
   to the base plate.
3. **Yaw head.** Drill the ring's bolt pattern into both 1/2 in plates. Drill the four 3/8 in
   drain holes. Bolt the ring to the fixed plate; bolt the rotating plate on top; turn it by
   hand and confirm it is silky before anything else goes on it.
4. **Drip hat.** Weld three small tabs to the fixed head plate and bolt the inverted 1 qt bowl
   (B3) over the ring, clear of the rotating plate.
5. **Nod yoke.** Cut the two cheeks from 1/4 in plate; bolt the two UCFL205-16 flange units (C1)
   to them; set the four urethane pads (C3) as stops, with shims, at the chosen lean. Bolt the
   yoke to the rotating plate.
6. **Boom.** Cut the 2-1/2 in tube to 144 in. Weld in the fulcrum cross-tubes and the pivot
   sleeve. Weld the counterweight stub (C2) coaxially into the butt. Weld the five 1-1/4 in
   collars (D3) and then the five stalks (D2) at their fan angles. Close the tube ends and drill
   the weep holes. Dry-fit the boom in the yoke; slide the two 25 lb plates (D4) on and set the
   balance so the arm floats just bloom-heavy.
7. **Blooms (five, identical).** Cut each bowl in half on its diameter and trim the rims so the
   two halves match. Cut the ten 3-1/4 in hub-cap discs from 1/4 in plate. Cut each hub from the
   1 in stainless tube, press the two 608-2RS bearings (E4) in, and fit the 8 mm axle (E3).
   Bolt each bowl half to the upper and lower hub caps, offset 1-1/2 in off the axis in opposite
   directions, using the M4 screws (F10). Spin each rotor by hand before it goes anywhere near
   the sculpture.
8. **Fit the blooms to the stalks.** Each bloom's axle drops into the stalk top and is held by
   one M4 screw. Rotate each successive stalk about the boom axis by the golden angle (137.5°)
   from the last, so the five blooms spiral around the arm instead of lining up.
9. **Finish.** Degrease and abrade the armature. **One single-component direct-to-metal enamel
   (F3) straight onto the prepared steel — no primer line** (decision 3). Charcoal-green on the
   armature and the bloom exteriors, clay-red inside the bowls (decision 1 — see §2.5 for the one
   open item about getting those colours out of a fixed-palette product). Mask the bearing bores;
   do not paint anything that has to fit. Boeshield (F12) the nod pin, the axles and the bare
   threads; 3M 5200 (F11) under the ring's bolt heads where stainless meets aluminium.
10. **Assemble on the pier.** Bolt the base plate down (A7). Lift the mast, then the head, then
    the boom. Re-set the counterweight last, on the finished piece.
11. **Commission.** Turn the ring by hand. Push each bloom and let go. Lean on the arm and let
    it go. Log the hand-spin time of each bloom as the baseline — that number is what tells you
    next year whether one has started to drag.

### 5.3 Schedule, and the one lead time that drives it

The sealed main ring (B1) ships in **25–30 working days — five to six calendar weeks — and it is
the critical-path item on this build.** That is a real constraint on the schedule, not a footnote:

- The ring must be **on order before the head plates can be drilled**, because the bolt pattern is
  taken from the igus CAD rather than measured. The CAD is available independently of the order,
  so the drawing work can proceed while the ring is in transit — but the drilling is not finished
  until the part is in hand to check against.
- Ordered in the first week of November, the ring lands **mid-December at best**, and year-end
  holidays push a realistic date into **early January**. Erecting an 11 ft 6 in steel piece on a
  12 in pier in a Grand County January is not a sensible plan, and the ground the pier goes into is
  frozen from about that point on.
- So the honest schedule is: **order the ring first**; dig and pour the pier before the ground
  freezes if the engineer's check allows it, otherwise in the spring; fabricate the armature over
  the winter; **erect in spring 2027.** Compressing that by switching back to the in-stock standard
  ring is precisely the trade Mike already declined in decision 2, and it is not re-opened here.
- And **nothing is ordered at all until the structural check from decision 5 comes back**, which
  puts the engineer's turnaround in front of the ring's.

---

## 6. The two risks the concept flagged, answered

### Risk 1 — it is a bearing machine: how the main thrust bearing survives dust and freeze-thaw

The main ring is an **igus iglide PRT-01 slewing ring** (B1). It is the one large bearing the
whole branch turns on, and it is a very specific choice:

- **There are no rolling elements and no grease.** A PRT ring is a hard-anodised aluminium ring
  running on replaceable iglidur J polymer sliding pads. There is no lubricant to hold grit, no
  lubricant to wash out, and none to stiffen in the cold. That removes the two things that
  actually kill garden bearings.
- **The material is rated for the climate.** iglidur J runs from −50 °C to +90 °C continuous
  (−58 °F to +194 °F), which brackets Grand County's −16 °F winter design temperature and the
  worst of an August in the sun, and the anodised aluminium housing cannot rust. The polymer is
  compliant, so a film of ice at the seal line is pushed aside rather than cracking a race.
- **Its failure mode is gentleness.** A worn pad is a slow loss of stiffness, not a seizure —
  and the pads are a replaceable part by design, which is the whole point of the PRT family.

Then the detail work, because a bearing is only as good as what is above it:

- The ring sits 7 ft 6 in up, so ground splash, grit and mower spray never reach it.
- An inverted 1 qt stainless bowl (B3) caps the ring and sheds rain and meltwater outward. This
  is the cheapest line in the whole bill and it is doing real work.
- Four 3/8 in drain holes in the fixed plate mean water cannot stand under the ring.
- Stainless bolts into the aluminium housing are bedded in 3M 5200 (F11) so the two metals are
  not in direct contact — galvanic corrosion outdoors is a bigger threat than the bolted load.

**Decided (Mike's decision 2, 2026-11-01): the fully sealed ring, PRT-01-100-DD.** It seals *both*
faces with sealing lips and is the version igus describes as built for "blowing sand… outdoor
mounts where both faces see weather", which is what this site is. Three things are recorded with it
rather than glossed:

- **The price is not published.** igus quotes the -DD variant on request. Until that quotation is in
  writing, this line carries a **floor allowance of $398.83** — the published price of the same-size
  standard ring — and the bill is stated both ways (§4). It is the only unpriced line left in the
  build, and it says so on its own line in the BOM.
- **The lead time is 25–30 working days** and Mike accepted it. It is now the critical path of the
  whole build schedule; see §5.3. That is the price of the seal, and it is a schedule price rather
  than a money price.
- **The bowl drip hat (B3) stays.** Sealing lips protect the sliding faces; they do not stop water
  standing on the fixed plate under the ring, which is what the four 3/8 in drain holes and the
  inverted bowl are there for. The $8.99 bowl is not made redundant by the $398 part above it.

**Sizing, and one honest gap.** The ring is sized by its permissible tilt moment, which is driven
by wind pressure on the five blooms acting about the mast axis. That calculation is not in this
sheet and must not be: the project brief for this piece is form and concept, and puts structural
engineering out of scope. What sits here is the *choice* with its reasoning — a 185 mm ring rather
than the 80 mm smallest size, because the arm is 12 ft and the blooms are 13 in across — plus, now,
the numbers the engineer needs to size it properly: the moving mass, its centre of gravity, the
wind-catch areas and their lever arms. Those are in `docs/slow-bloom-structural-brief.md`. The
service-life computation is a separate piece of work and should be run through igus' PRT online
configurator before the drawing stage.

### Risk 1b — every other place water can do damage

A bearing is not the only thing a freeze can break; a **closed steel tube full of water** is. So:

- Every tube end gets a plastic plug (D5, D6) and every closed volume gets a 3/16 in weep hole at
  its lowest point.
- The bloom hub is stainless (E2) and drains from its underside.
- The head is not a closed box — the two plates are spaced and open at the sides.

### Risk 2 — it is a bearing machine: how the per-bloom bearings survive, and why one sticky bloom does not read as broken

Each bloom spins on **two 608-2RS sealed bearings** (E4) pressed into a stainless hub. 2RS means
a rubber contact seal on both faces — they exclude dust and blow-off grit and hold a sealed-for-
life grease. They are the most common sealed bearing in the world.

- **Dust and grit:** the seals do the work; the hub is stainless, so even a failed seal leaks
  water onto stainless, not onto rust.
- **Freeze-thaw:** the real hazard is water *standing* in the hub. So the axle is vertical with a
  hat above the top bearing and a drain below the bottom one, and the lower bearing gets a
  pressed-in shield. The bearing's own grease stiffens with cold, which raises the torque needed
  to start — but the rotor is a drag machine with surplus torque at low speed, so cold grease
  costs a little spin, not a stall. A two-lobe offset rotor does not need a threshold wind speed
  to start from rest; that is exactly why it was chosen over a propeller.
- **Replaceability is the actual answer.** The rotor lifts off its axle after two M4 screws. A 608
  costs $1.50 in tens, and the bill buys twenty of them — ten in service and ten on the shelf.
  Turning a stuck bloom into a working one is a ten-minute job done standing on the ground.

**Why one stuck bloom does not make the piece look broken.** This is answered by the design, not
by the bearing:

- **No single bloom carries the read.** The primary motion — the grand slow turn of the whole
  branch, and the lean — is carried by the main ring (B1) and is completely independent of every
  bloom. If all five blooms seized solid, the piece would still turn, still lean, still read the
  wind, and still be worth looking at. That is the test this design has to pass, and it passes
  it. The blooms are the flicker; they are not the subject. Contrast the Wind Curtain direction,
  which the concept sheet rejected partly because it makes every element load-bearing for the
  read — hundreds of chances to look broken.
- **The blooms are built to start, not to be coaxed.** Low-inertia 0.4 mm bowls on sealed
  bearings, driven as drag rotors, turn in a breeze that a propeller-driven ornament would
  ignore.
- **A service plan that fits a garden.** Once a year: turn the ring by hand, push each bloom and
  compare its time against the commissioning baseline, check the nod pin, one grease-gun stroke
  in each flange unit. If a number is off, swap that bloom's bearings. That is the entire
  maintenance schedule, and it is why the bill includes a spares kit rather than a warranty.

### Risk 3 — why one pivot too many does not make it a Calder knockoff

The concept sheet named this as the subtler risk, and it is a design argument rather than a
mechanical one. Three differences, in order of importance:

1. **Nothing hangs.** Every element is a stiff cantilever: the boom is welded to the head, the
   stalks are welded to the boom, the blooms are bolted to their axles. A Calder mobile is
   *defined* by suspension — weights on wire, whose balance is never settled, always drifting.
   Slow Bloom's arm is held and driven; it is not a mobile because there is nothing free about
   it.
2. **The lean is against a spring, not about a balance point.** This is the mechanical heart of
   the answer. A mobile's arm swings freely about its balance and can oscillate; this arm tips
   3–8° into a gust, is stopped and pushed back by four urethane pads, and returns to exactly the
   attitude it left. It cannot oscillate. That single difference in mechanism is what makes the
   piece a planted instrument rather than a Calder.
3. **The vocabulary is botanical, and it is small.** Five blooms, one stalk, one straight bar, no
   coloured discs, no wire, no dangling hardware, and no primary-colour assortment — the whole
   piece is two colours and stainless. Five shallow split bowls read as a flower because they are
   arranged like one: staggered, spiralling at the golden angle, no two the same height.
   Restraint is the argument. One stalk, five blooms, three motions, nothing else.

And it is honest about its ancestor. The reference sheet says plainly that the piece is a Calder
standing mobile **made literal and planted**, and that it departs from Calder at the drive — the
wind sets its direction rather than merely disturbing it.

### Risk 4 — a genuine gap, now handed to the person who can close it

This sheet is a form document. It contains no structural calculation: the pier, mast and the
115 mph design wind are named but not resolved. That is consistent with the project brief (form and
concept over structural engineering), but it is a real limit of this deliverable rather than a
design decision.

**Mike's decision 5, 2026-11-01: the check gets done, before any metal is cut.** It is not done
here and it will not be done here — **structural engineering is outside this project's scope and it
is a licensed professional's job.** Grand County's own code document notes that a licensed Colorado
design professional may be required for structures in special wind regions, and this county has
them. What has been produced instead is the package the engineer needs:
**`docs/slow-bloom-structural-brief.md`** — the geometry as drawn, the itemised moving mass with its
centre of gravity, the wind-catch areas and their lever arms, the design wind, the foundation
assumptions, the questions to put to him, and a plain statement of what is not known. Its numbers
are computed from `output/slow-bloom-load-inputs.json` by `scripts/slow_bloom_loads.py`, so they can
be checked line by line instead of taken on trust.

The reason it matters, stated once: the itemised moving mass comes to **about 155 lb**, not the
"~90 lb" this sheet claimed at NDY-2, and the wind figures are large enough that an order-of-
magnitude pass puts the overturning moment at the base in the low thousands of foot-pounds. Those
numbers are inputs, not conclusions — but they say plainly that this is not a piece that can be
planted on an assumed foundation.

---

## 7. Mike's five decisions, recorded — and what is still open

All five were settled on 2026-11-01 and are applied in this revision. They are recorded here as
outcomes rather than re-opened as questions.

1. **Colours — pairing 1.** Charcoal-green stem, marigold blooms, clay-red interiors. Applied in
   §2.5 and §5.2 step 9.
2. **Main ring — the fully sealed PRT-01-100-DD.** Applied to the BOM at line B1, with the price
   carried as a floor allowance because igus publishes none, and the 25–30 working day lead time
   written into the schedule at §5.3.
3. **Finish — the single-can direct-to-metal enamel at $53.98.** Applied: F1 and F2 deleted, F3
   promoted to primary, total recomputed, and the shorter service life stated in §2.5 rather than
   glossed.
4. **Blooms — the thin 0.4 mm bowl at $17.29.** Applied at line E1, with the dent-and-replace trade
   recorded on the line itself.
5. **Structural check — yes, before any metal is cut.** Applied as `docs/slow-bloom-structural-brief.md`.
   The check itself has not been done here and will not be; it is a licensed professional's job.

**Still open — three items, all named rather than hidden:**

1. **The ring's actual price.** igus has not quoted it. The bill carries a $398.83 floor allowance
   and gives the total both ways (§4). Everything else in the bill now has a real price.
2. **The coating colours.** Decision 1 wants three colours; decision 3 bought a product that cannot
   be tinted. §2.5 sets out the three ways out; the bill carries the flat-black gallon as the
   decision's price basis until it is settled.
3. **The structural check itself.** Awaiting a licensed engineer. Nothing is ordered, cut or welded
   before it comes back.

## 8. Verification

- The updated BOM was run through `scripts/slow_bloom_bom.py` on 2026-11-01:
  `python3 scripts/slow_bloom_bom.py --file docs/slow-bloom-bom.json` exits 0 and reports
  `validation.ok = true`, with no missing part numbers, no missing prices, no missing links, no
  placeholder text and no duplicate line ids. **36 lines, all 36 primary, $2,396.48** — against
  $2,832.48 before the decisions, a saving of **$436.00**. The tool's one warning is the honest one:
  `listing=5, quote=1` — six lines whose price was not read off the vendor's own product page.
- The structural input data was run through `scripts/slow_bloom_loads.py` on 2026-11-01:
  `python3 scripts/slow_bloom_loads.py --file output/slow-bloom-load-inputs.json` exits 0 with
  `validation.ok = true`. It reports 156.53 lb of rotating mass (of which 50 lb is counterweight),
  127.76 lb fixed, a dead-load imbalance of 58.9 ft·lb bloom-heavy, only 20.8 ft·lb of counterweight
  trim authority, 10.08 sq ft of wind-catch area, qz = 28.8 psf, and an indicative base overturning
  moment of 2,011–4,022 ft·lb. **Those are inputs, not design values, and the tool says so in its own
  output.**
- 30 of the 36 prices were read directly off the vendor's own product page on 2026-11-01 and are
  flagged `page` in `docs/slow-bloom-bom.json`; 5 are flagged `listing` (A1, A2, C1, F3, F11) and
  each carries a note saying exactly where the price came from and what the alternatives cost; 1 is
  flagged `quote` (B1) and carries the allowance basis in full.
- Two parts that genuinely cannot be sourced with a published price are named explicitly rather
  than faked: the 14 in form tube (which is why the pier is 12 in) and the fully sealed
  PRT-01-100-DD ring — the latter now being the chosen part, which is why the total is stated twice.
- Two arithmetic corrections were made to this sheet in this revision rather than carried forward:
  the moving mass (about 90 lb → about 155 lb, itemised and computed) and the pier depth (the 3 ft
  form tube with 2 in proud left 34 in in the ground, 2 in short of the adopted 36 in frost depth;
  the tube is now cut to 38 in).
- **Nothing in this sheet has been ordered, cut, welded or paid for.**
