Three kinds of number

“List price” is doing three different jobs in this file and they carry three different weights of evidence. Every row below states which one it is, and the figure itself is drawn differently in each case.

The rule that picks the figure is the operator's, and the file states it: “LIST IS THE HIGHEST PRICE OBSERVED across sellers.” Where sellers disagree the gap is a promotion someone is running today, not a second opinion about the list.

One record classifies differently by hand than by rule. The hand wins — a person read the page and this code did not — and the disagreement is printed rather than resolved: viltrox_af_23mm_f1_4_xf is filed “observed — maker store only”, where the sources alone would make it highest observed.

The 4 prices nobody published

These are arithmetic. A seller showed a selling price with no struck-through list beside it, and the cents gave the promotion away — a figure ending .20 or .25 is a percentage off something rounder. Reversing the percentage recovers a list that no page states. They are printed in red under a dashed rule for the whole of the rest of this page, because that is what they are.

Does the money predict the grade?

The rubric is not allowed to look. price, discount, value_for_money, brand_origin, country_of_manufacture are all named in the ledger's excluded factors, so no grade on this page was reached by consulting a price. That leaves the separate, answerable question of whether it would have worked — and across 99 graded X-mount lenses from $59 to $3,999.95 (median $579), Spearman's rho is -0.716. Negative because the grade scale ascends as quality falls: dearer really does mean better, strongly.

With the tie correction, which this one needs. 99 lenses share only 6 grade values, so the grade ranks are almost nothing but ties — and the schoolbook 1 − 6Σd²/n(n²−1) shortcut, which is Spearman's rho only when nothing ties, stops having a single answer here. Break the ties one way and it reads -0.573; break them the other and it reads -0.829, with nothing in the data preferring either. That interval is the argument for the correction: a figure that moves by 0.256 depending on an arbitrary choice is not reporting the data. The corrected value — -0.716, the correlation of the tied mid-ranks — does not move.

By price band

That is the finding, and it is two claims rather than one. Between tiers the money sorts correctly. Nothing below $168 grades better than B+ — the cheapest lens in the ledger at A− or above is the TTArtisan 27mm F2.8 AF. Nothing dearer than $999.95 grades below A+.

Inside a tier it sorts nothing at all. The 2 middle bands hold 5 different grades apiece on the same money, and the boundary price above is itself the finding in miniature: 3 lenses sit at exactly $999.95 and grade A, A-, B+. A cheap lens is not an expensive lens; a $579 lens could be either.

What one grade costs

The same question from the other side. Fix the grade, and see what the market charges for it.

Every grade in the ledger spans at least 2.5× from its cheapest member to its dearest, and the widest runs 15.6×. Two lenses the rubric cannot tell apart can be a factor of 16 apart in money. That is the same finding as the band table, restated in the direction a reader actually shops in.

One maker, two badges

Samyang and Rokinon are one manufacturer's output sold under two names, which is why the ledger carries 6 merged records rather than 12 listings. The seller stocks both badges, so for 2 of those records the pair can be priced against itself — and where both names appear here, they agree exactly. Off our ledger they do not.

THE SAME LENS COSTS DIFFERENT MONEY UNDER ITS TWO NAMES. Samyang and Rokinon are one manufacturer's output sold under two brands — which is why the ledger merges them into a single Samyang/Rokinon record — and B&H stocks both, so the pair can be priced against itself. Where the names agree they agree exactly: 135mm f/2 $549, 85mm f/1.4 $299, 8mm f/2.8 Fisheye II $299, AF 75mm f/1.8 $449, 12mm f/2 AF $499. Where they disagree the gap is large and one-directional — the 14mm f/2.8 Mk2 is $399 as a Rokinon and $499 as a Samyang, the 100mm f/2.8 Macro $499 against $549, the 16mm f/2 $315 against $359, the Reflex 300mm f/6.3 $209 against $269. Identical optics, up to $100 apart, decided by which badge the box carries. The merged record is the honest one; a ledger that split them would publish two prices for one lens and call it two products. hand-prices.json · data findings

Where the sellers disagree

19 of the 132 records have two quotes that do not match, and the disagreement has one shape. In all 19 the maker's own store is the lower figure, and in all 19 it says “as low as” rather than naming a price — a promotion marker, not a list. So the gap below is today's discount rather than a second opinion, which is what licenses the rule of taking the highest quote. $3,750 of markdown across 19 rows, all of it on one seller's side of the disagreement.

What has no price

52 of the 152 X-mount records carry no figure, because no seller in this file stocks them. That is an em dash, not a zero and not a free lens — the ledger's rule everywhere is that an absent value is not a negative value.

The M-mount side is thinner still and the shape of the thinness matters more than the count. 32 of 100 records are priced, all 1 of the ledger's 6 brands, all from CameraQuest. So the M-mount figures are one specialist's catalogue rather than a survey of the mount, and they cannot join the correlation above in any case: the M ledger leaves every grade null on purpose, because a grade is editorial judgement and a harvester producing one would be inventing the opinion.

Every price

All 132, dearest first, each with the basis it rests on. observed · highest observed ≈ · derived †.