Data and sources.

Every trade constant this site publishes carries its origin, the date it was last checked, and one of three statuses. The distinction is not bureaucratic: a figure fixed by a standard, a figure read off a shelf last month, and a figure everybody repeats without a source are three different kinds of claim, and treating them alike is how calculators end up asserting folklore. The register holds home improvement constants only — board, roll, panel and tile sizes, allowances and installation checks — because home improvement is the only subject Calculiva covers.

The three statuses

Standard

Taken from a published reference document that can be cited by title and section — a product standard, a federal handbook or a published reference table — and cited with its reference. It does not mean the value is a legal requirement.

Market survey

Read on products on sale or on tools as they ran, on the stated date. Every one carries an expiry date.

Convention

Widely repeated, with no normative source. Published as a convention, never as a rule. A convention may carry an expiry date too.

The register

Download the register

Calculiva’s contribution to the register above is published as open data under CC BY 4.0: the table as CSV, the records as JSON — with their values where the publisher’s terms allow republication, otherwise with the source to read them at — and a data package descriptor with the schema. 35 records, last verified 2026-09-25. Reuse it with attribution to Calculiva and a link to this page. The license covers Calculiva’s own contribution only — the compilation, its structure, the statuses, scopes and notes. Text quoted from a source, and the charts, list prices, product dimensions and tool outputs read from one, stay attributed to that source and dated; no right over them is granted or claimed, and Calculiva is not affiliated with any company named.

Studies

Plank waste study: engine-counted offcuts vs flat waste allowances

Version 1.0.0, dated 2026-10-02. License CC BY 4.0.

The Calculiva flooring layout engine planned every rectangular room from 72 to 288 in (6 to 24 ft) on each side, in 6 in steps — 1,369 rooms — for six plank formats laid with rows staggered by half or a third of a plank, and counted the whole planks each plan consumes. That count was set against what a flat 5, 8, 10 or 15 % allowance orders for the same room. “Reuse” pairs a row’s start piece with an end piece cut from the opposite end of the same plank; “no reuse” throws every end offcut away, a worst case rather than an installation practice. The four allowances are comparison levels, not recommendations.

• Offset 0 (no stagger) was also run on the three sourced formats, as a reference only: it is not an installation pattern and it is excluded from every headline figure.
• The parametric formats (36, 60 and 72 in long at 7.5 in wide) isolate the effect of plank length. They are not retail formats, and no product is claimed to exist at these dimensions.
• Kerf 0 is a floor; 3 mm is a sensitivity value, not a statement about any saw blade. Rooms are empty rectangles with zero clearance, one plank per carton and no reserve.

Reading the first line: on the sourced formats, with reuse and no kerf, a flat 10 % allowance came up short on 16.8 % of the 8214 plans and ordered more than needed on 76.9 %. The figures for each format and offset, including the reference-only offset 0, are in the summary file.

Files: README (method, parameters, limits and four cases checked by hand), rows.csv (one row per plan), summary.csv (one row per format, offset and scenario) and the data package descriptor. Every number in them was computed by the engine; no third-party waste figure enters the calculation, so the CC BY 4.0 license covers the files whole. The engine code is not published; any row can be rerun on the plank layout calculator with the settings the README lists, comparing its “Source boards” result.

The same four files, byte for byte, are archived on Harvard Dataverse under a persistent identifier: doi:10.7910/DVN/8VL3MW. Cite that DOI when a citation needs to outlive this page.

What expiry means here

Every market survey carries an expiry date, and a convention may carry one. Past that date the record is left out of the dataset at the next export, and the site build stops until its source has been read again: an expired figure is not shown with an old date. A price that was true a year ago, displayed today without saying so, is worse than no price.

What we deliberately do not publish

• Saw kerf. It belongs to your blade, not to a table. It is an input on every page that needs it, with a starting value you are told to replace.
• Market prices. No average, no regional figure. The only prices read from a seller are brand list prices, quoted with the date they were read and dropped when their record expires: the York and Painted Paper wallpaper pages carry them. Every other price field is a quote you enter, and its starting value is an exercise, not a market figure.
• Brand product data we have not read ourselves. A page named after a retailer that does not carry that retailer's real roll widths, box coverages or panel sizes has no reason to exist.

The rounding rules and the limits of the geometry are on how we calculate.