Field Notes
Eight things the catalogue turned out to know. Every figure below is computed from the 693 records themselves — the timestamps, the coordinates, and what each object turned out to be.
Half the gloves arrive in two months
Share of each category found in each month. Gloves are stacked into deep winter; peels are spread across the year and peak in March.
49.6% of all gloves were found in December or January. Across the six months from May to October there are 43 gloves — fewer than in a single December.
A winter spike on its own would prove nothing, since it could just reflect a collector who walked more in the cold. The peels rule that out. Same person, same streets, same camera, and they peak three months later. The two distributions come apart, so the glove pattern is weather rather than habit.
The gloves found in summer are a different kind of glove
In cold weather the gloves people lose are the ones they were wearing to keep warm. By July most belong to somebody working outdoors. The object does not just become rarer in summer, it becomes a different object. These summer figures rest on 43 gloves, so they describe a small population.
In April 2020, half the gloves were disposable
Share of each year's gloves that were nitrile, latex or vinyl.
The Netherlands locked down on 15 March 2020. Disposables ran at 5–7% of finds through 2018 and 2019, reached 30% that March and 51% in April. They never return to the old baseline: 2021, 2022 and 2023 all settle around 12%, roughly double what came before.
Counting gloves could not have found this. 2020 was the biggest year for gloves and peels alike, which is the signature of somebody walking more during lockdown rather than of more gloves reaching the ground. Only the composition shows it.
Gloves arrive in runs
Finding a glove is not an independent event. Shuffle the locations while keeping every timestamp and you would expect about 22 closely spaced pairs; the catalogue has 132. The best day in nine years is 1 January 2018 — 7 gloves on one walk.
Peels are found among rubbish. Gloves are found in the road.
Both objects end up mostly on pavement, as anything dropped in a city does. The difference is in the tails. Peels are 7.0 times more likely to be lying among other litter — beside bins, in tree pits, on ledges, in the general vicinity of where rubbish goes. Gloves are 4.6 times more likely to be in the roadway.
That asymmetry is the whole difference between the two categories, stated without guessing at anybody's intentions. A peel is put down by the person who finished the banana, somewhere that felt tidier than the pavement. A glove goes wherever it fell, including into traffic, where nothing is ever placed on purpose.
There is no such thing as a slightly old banana peel
Peels were graded on a seven-stage scale from fresh yellow to a pigment stain with no form left. The distribution has two humps and a dip between them: they are found either fresh or well rotted, and stage III passes quickly enough that it is rarely what you walk past.
Stage VII is unattested. 201 peels and not one ghost — either the final stage does not survive as a photographable object, or it was never once encountered.
The pavement is a thermometer
Every find matched to that day's Amsterdam temperature, then read as a proportion: of everything found on days in each temperature band, what share was a glove?
The slope runs the whole way without a single step out of place. Below freezing, 94% of what was on the ground was a glove. Above 20°, 21%. The median glove was found at 6.7°, the median peel at 8.8°. The two ends of that range are thin — 17 finds below zero and 14 above twenty — so it is the middle four bands, 655 finds between them, that carry the slope.
Reading it as a share rather than a count is what makes it mean anything. The number of objects found on a given day depends on how far somebody walked; the ratio between the two categories does not.
And it is not just winter in disguise. Take only December, January and February, so the season is held still, and split those days at their own median temperature of 5.4°. The colder half still runs 90% gloves against 80% on the milder half — a gap that would arise by chance about 0.9% of the time. Cold does its own work on top of the calendar.
Rain does nothing at all. 71% on dry days, 71% on wet ones. Whatever puts gloves on the pavement, it is temperature and not weather in general.
One last thing the dates give away. In 6 of the 7 winters, the first knitted glove of the season appears before the first night of frost — a median of 15 days before. People do not dig out their gloves when it freezes. They dig them out when it merely starts to feel like it might.
Winter is knitted. Summer is rubber.
Every glove classified a second time, by what it is made of rather than what it is for. 492 gloves, seven materials, judged from the photograph.
Split that by season and the two halves of the year barely share a material.
In the cold months almost half of what turns up is knitted wool or acrylic. In the warm months half of it is coated work glove — the rubberised, dipped kind somebody was wearing to do a job. Fleece disappears from the record entirely between May and October. Not a single one.
The materials sort themselves by temperature, which nothing about the labelling forced them to do. The median leather & suede glove was found at 5.5°, the median coated & rubberised glove at 8.8° — more than three degrees apart, on a scale where the whole collection sits within a few degrees of freezing.
One thing this exposes about the catalogue itself. The genus Nivalis is described as the winter knit, and only 61% of the gloves in it are actually knitted: the rest are leather, fleece and synthetic shell. What a glove is for and what it is made of are different questions, and this archive had only been answering the first one.