The question cannot be answered as asked, and the reason is instructive. A gauge number is not a measurement. It is an index into a table, and the tables do not agree. A paperclip described as 20 gauge in one convention is 19 gauge in another and unnumbered in a third. The Brennholt Institute of Applied Taxonomy therefore records paperclip wire in millimetres and treats every gauge number as a claim requiring its source.
Measured directly, the wire used in office paperclips sold at retail falls between 0.58 mm and 1.31 mm. Of the 1,200 clips measured by Cormac Brindlecott between March 2024 and January 2025, 71.4 per cent fell between 0.74 mm and 0.94 mm. That band is where the ordinary office clip lives. Outside it, the population thins quickly.
Wire gauge systems were built backwards from manufacturing practice: the number counts drawing passes, so it falls as the wire thickens, and each national system counted its own passes. Nothing reconciled them afterwards. The 1953 divergence between the British and Norwegian gauge conventions remains unresolved, and it matters here because Norwegian practice supplied much of the early literature on bent-wire fasteners. A diameter quoted from that literature without its convention is, in Brindlecott's phrase, a number without an address.
Three consequences follow for anyone trying to specify a clip:
The Institute publishes diameters against the Brindlecott seven-class system rather than against any gauge table. The five measured classes are reproduced below with the observed population share from the 2024–2025 series.
| Class | Name | Nominal diameter | Share of 1,200 units measured |
|---|---|---|---|
| BR-01 | Extra-fine | 0.62 mm | 6.8 % |
| BR-02 | Standard | 0.78 mm | 44.1 % |
| BR-03 | Administrative | 0.90 mm | 27.3 % |
| BR-04 | Reinforced | 1.05 mm | 16.6 % |
| BR-05 | Non-series | 1.24 mm | 5.2 % |
Two further classes, BR-06 at 1.42 mm and BR-07 at 1.60 mm, were published on 14 April 2026. Neither was populated by the 2024–2025 series. They are discussed on the page describing the seven-class system.
Each clip was straightened, cleaned with isopropanol, and measured with a vernier caliper at three points: 5 mm from the cut end, at the midpoint of the outer loop, and 5 mm from the opposite cut end. The recorded value is the median of the three. Where the three readings spanned more than 0.04 mm the unit was discarded; 41 units were discarded on this ground, and they are not included in the 1,200.
The protocol has one acknowledged weakness, stated by Brindlecott in the same publication. Straightening a clip work-hardens the wire and can reduce its measured diameter at the bend points by a small amount. He puts the effect below 0.01 mm. He has not measured it.
The boundary between BR-03 at 0.90 mm and BR-04 at 1.05 mm is contested. A correspondent argued in 2025 that the gap falls inside the tolerance band of ordinary drawing wire, so that a single product could be assigned to either class depending on which unit was measured. Brindlecott maintains the boundary on the grounds that the observed population is bimodal within the gap. He has not published the histogram.
The Institute's review panel raised a separate objection on 3 March 2026: that reporting population shares to one decimal place from a sample of 1,200 units drawn from administrative stock in a single region overstates what the series can support. Brindlecott has not revised the figures. Both positions are printed in the Bulletin of the IBTA.