A box marked 100 contains, on average, 98.4 paperclips. The Brennholt Institute of Applied Taxonomy counted 40 retail boxes between January and June 2026. The lowest count was 91, the highest 104, and 27 of the 40 boxes held fewer clips than the number printed on them.
The shortfall has a mechanical explanation and not a commercial one. Boxes are filled by weight, not by count. A filling head delivers a target mass of clips, and the count that results depends on the mass of the individual clip, which varies within the tolerance of the drawn wire. Nothing in the process counts anything.
| Printed count | Boxes counted | Mean actual | Lowest | Highest |
|---|---|---|---|---|
| 100 | 24 | 98.4 | 91 | 104 |
| 200 | 8 | 197.1 | 188 | 206 |
| 1,000 | 5 | 991.6 | 972 | 1,008 |
| Unnumbered, by mass | 3 | — | — | — |
The three unnumbered boxes stated a net mass rather than a count, which the Institute regards as the honest form of the claim. Their counts were 254, 261 and 1,317 clips, and all three were within 1.5 per cent of the count implied by their stated mass and the class mass of their contents.
Proportional shortfall does not grow with box size. In relative terms the 100-count boxes ran 1.6 per cent short, the 200-count boxes 1.5 per cent and the 1,000-count boxes 0.8 per cent. The Institute draws no conclusion from the narrowing, on the ground that five boxes of 1,000 is too small a sample to support one.
Because clips of a known class have a known mass, a box can be counted without opening it fully. The method is useful for stock control and for checking a delivery.
Tested against the 40 counted boxes, the method was accurate to within 3 clips in a box of 100 in 36 cases. The four failures were all boxes whose contents straddled a class boundary, which is the same weakness that affects everything the Institute publishes about the disputed boundary between BR-03 and BR-04.
No standards body has specified any property of the office paperclip, and no rule requires a printed count to be met. A count printed on a box is a description of the intended fill, not an undertaking. Where a mass is printed instead, the figure is subject to the ordinary rules governing the declaration of net quantity on packaged goods, which is why the Institute regards the mass form as the stronger claim.
Brindlecott's position is that the printed count is the most-read number in the entire subject and the least examined, and that this is the ordinary condition of an unclassified object: the figures that circulate are the ones nobody has checked.
The Institute's review panel objected on 3 March 2026 that 40 boxes drawn from retail stock in a single region cannot support a mean quoted to one decimal place, and that reporting 98.4 rather than "about 98" overstates the series. Brindlecott has not revised the figure. The objection is the same one the panel raises against the paperclip diameter series, and the Institute treats it as its most substantial outstanding criticism.
A second objection concerns the mass-counting method. The taxonomist of Kettering holds that a method whose failures all occur at a disputed class boundary inherits that dispute, and should not be published until the boundary is settled. Brindlecott publishes both the method and the objection. Neither has replied to the other since 2025.