How the drawings were measured

The ruler

The July 1912 advertisement is filed twice, as a page render and as a native-resolution crop of the same printing, and the two score 0.975. That is what the same artwork looks like as a measurement, and nothing else in the collection reaches it.

Two independent digitisations of the same July 1912 advertisement. Their score is the noise floor: no pair of genuinely different drawings should be read as 'the same artwork' above it without other evidence.

Measured, not described. Each drawing is reduced to the same thing — the letters BEND, cropped to their own ink — and compared with every other by shape. Then every pair is measured again at four different widths of the ring, because a score that moves when the ring moves is a fact about the ring.

Each pair is measured again at 4 device cuts — 0.96, 0.9, 0.84, 0.78 of the roundel’s interior — and a pair counts as stable only if its score moves less than 0.05 across all four. Of 1081 pairs, 170 hold at all four.

The two groups that hold

The drawing the club used, and what holds with it. The 1912 and 1916 newspaper impressions score 0.930 against each other and barely move when the device is cut differently — the same printing cut, four years apart, just below the 0.975 that two scans of one printing reach. The Bulletin’s own business directory masthead of February 1939 joins them at 0.887 against 1912 and 0.861 against 1916, holding within 0.009 and 0.023 across every cut: the tightest reading anything between 1917 and 2001 has produced, and still short of the same artwork. The museum’s June 2020 drawing sits at about 0.69 against 1912 and 1916, steadily. This is a chain rather than a circle, and the shape of it is worth saying plainly: 1939 and the museum’s 2020 drawing hold with the two period impressions but not with each other — that pair runs 0.612 to 0.700 and crosses the grouping threshold depending on where the device is cut. They are shown together because each of them holds with the 1912 cut, which is what this group claims and all it claims. Bend Park Company, July 1912 · The Bend Bulletin, November 1916 · Bulletin business directory, February 1939 · Deschutes Historical Museum, June 2020

Two files, one artwork. The gold roundel on Downtown Bend’s history page and the museum’s February 2012 artwork score 0.962 against each other and hold that score at every device cut. Two files of one drawing, and the museum is the earlier known holder of it. A third file sat here until 2026-09-14: a gold image found on the internet in March 2022, which measures with these two at 0.954 to 0.959. It is now kept as a reference rather than as a specimen, because where it was found says nothing about where the drawing came from, and a comparison group is a claim about the collection. Downtown Bend gold roundel · Deschutes Historical Museum, 2012

A high steady score says two drawings have the same shape. It does not say who copied whom, or which came first.

Start with the pair the ruler was made for — the highest-scoring distinct pair among the drawings the board plates, at 0.962, which reads as “very close; the same drawing, or one traced from it”.

Measured, and standing alone. Each of these was measured and joins no stable cluster at the global cut. They are not a group; they are 32 separate drawings shown together because that is what one of the two measurements says. Read the character matrix for what each one is made of, and the letter cut for how alike a pair is once the ring is out of the way. Eight were added on 16 September 2026 from the Meta sweep, and not one of them made a stable pair at or above the threshold with anything already here — including the City’s own current mark against the City’s own 2023 neighborhood artwork, which the character matrix places in the same family. Two more were added the same day from the outsized-issue sweep: the Bulletin’s directory masthead as printed on 1 September 1943 and on 24 August 1949. Both are the same block as the 1939 plate, which the character matrix reads identically in all twelve states, and neither scores above 0.52 against anything on file — the 1939 plate included, at 0.557 on the letters. That disagreement is the record: ink weight is what the shape measurement cannot see past, and this masthead is the drawing the handbook already names for it. The count in this sentence had said 29 while the list held 30; it is counted from the list now.

11 drawings could not be measured from the images on hand. Each says why in its caption on the board and in full on its record.

What the frames on the board mean

A dashed frame is a drawing seen in use whose own date is unknown. A gold frame is a weak reading, and where the drawing is small enough that its size is the reason, the number beside its year is how many pixels wide one letter is. A ring means that drawing is tied to the one you picked, in that tie's own class.

What one picture on the board stands for

The board shows 47 pictures of the 58 drawings on file: one for every drawing, and one more for every distinct artwork made from it that the ledger has a sighting of — the letters stitched into knit, cut out of wood, set inside a ring of type. Each picture is the best of its own artwork: a picture at all, then the highest matrix read, then the earliest dated, then the largest device.

A family stops at twelve pictures. Above twelve, the holder with the most in that family loses its latest, one at a time, until twelve remain — a rule about keeping a variety of sources on the page rather than about which picture is best. The bureau roundel lost the most, 6 of them. Nothing is dropped from the archive: what the cap takes off the board is listed in a caption on it, each with its own picture, its own link and its own pick, and the line under the family says how many there were. The deepest caption is City of Bend in the bureau roundel, 6 behind one picture.

What “seen in use” counts

The archive files 65 documented appearances of the emblem, and 47 of them are drawings on the board. The rest are records that name the emblem without carrying a drawing this project can measure — a report of a device, a design brief's own words, a mark the emblem was replaced by — and the 2022 accession, whose two files are in the references below under their own ids. The board's filter counts what it shows, which is one drawing per holder per family: one counts when the archive files any of that holder's drawings in that family as an appearance, or the appearance ledger holds a dated sighting of one.