The 8.9% tip that the iceberg drew as 1.3%
Showing the misleading chart
A docents’ newsletter we drew shows The Met’s collection as an iceberg: 44,855 of 503,207 objects on view, 8.9%, above the waterline. The labels are right, but the waterline sits 8.9% of the way down the iceberg’s height, and an iceberg narrows to a peak, so the tip gets 1.3% of the ink. The drawing shows 73 parts in storage for every part on view; the catalogue says about 10.
01The claim
What visitors see at The Met is a sliver: almost the entire collection is hidden below the waterline, out of sight.
02The trick
Both labels are right. The Met’s open-access catalogue lists 503,207 objects, and 44,855 of them, 8.9%, have a gallery number, so 458,352 are in storage. The trouble is where the waterline was drawn: 8.9% of the way down from the peak, as if the share were a height. An iceberg narrows to a point, so the top 8.9% of its height is a thin spike that holds far less than 8.9% of its ink. On this outline the tip is 1.3% of the drawn area. For a plain triangle it would be 0.8%, because when a shape narrows to a point both the width and the height of the tip shrink together, so its area goes with the square of the height: 0.089 × 0.089. It is the square law behind every area illusion, run in reverse: instead of a doubled value growing four times the ink, a small share cut from a peak shrinks to a fraction of its own size. The eye compares ink, and the ink says 73 parts in storage for every part on view. The catalogue says 10.2. Anything drawn as the tip of a pointed shape gets the same treatment, a pyramid or a mountain as much as an iceberg. The smaller the share, the more it shrinks. (We drew this newsletter and its iceberg ourselves, as a demonstration; every count is The Met’s.)
03The fix
The redraw keeps the iceberg but places the waterline by area: to give the tip 8.9% of the ink it has to sit 23.9% of the way down, and the peak becomes a visible cap, one part in 11 rather than one in 74. Beside it, a plain bar chart gives each department’s share on view, which an iceberg cannot show at all. The spread is wide, with one caution: “on view” here means a gallery number in the dataset’s snapshot, and some galleries are dense study displays (4,526 Egyptian objects are recorded in one gallery alone). Egyptian Art has 73.5% of its objects in a gallery, European Paintings 43.0%, Drawings and Prints 0.04%, 79 of 185,361. Most of the storage is light-sensitive material. Drawings and Prints, Photographs and the Costume Institute together hold 51.0% of the catalogue, and only 349 of their 256,861 objects are on view, 0.14%. The Met shows works on paper in rotations of about thirteen weeks because light damages them, and museums ration light for photographs and textiles too, so much of this storage is the dark these objects need, not a failure to show them. Everything else in the catalogue is 18.1% on view, 44,506 of 246,346. If a share has to go on a pictured shape, cut the shape by area, not by height; otherwise use a bar, where length is the only thing to compare.