Base du système métrique · vol. 3 · the unit of weight, 1799
The brass cylinder that fixed the kilogram in 1799
The kilogram was to be the mass of a cubic decimetre of distilled water at its densest. Louis Lefèvre-Gineau and Giovanni Fabbroni found it by weighing: a hollow brass cylinder, measured in old French lines, was weighed in water kept near freezing and again in air. Johann Georg Trallès reported every step for the commission. Switch the corrections on and off and watch the result move.
18,827.15 grains, poids de marc
- Volume in the cold water, dm³ (recomputed)
- One dm³ of water, in the working unit (recomputed)
A cylinder measured in lines
No vessel of exactly one cubic decimetre could be made and filled accurately enough, so the experiment worked by displacement: weigh a solid body in air and in water, and the difference is the mass of the water it pushes aside. Lefèvre-Gineau had a hollow brass cylinder made, about 24 centimetres tall and wide, and measured it with rules compared on the instrument used for the metre’s base lines. Its height came to 108.061032 lines and its mean diameter to 107.6486 lines, the line being a twelfth of the old French inch; with the metre at 443.296 lines, those round to 0.2437672 m and 0.2428368 m. Its external volume at 17.6 °C is 11.29 cubic decimetres.
The report also prints the rule used for the diameter, 108.64658 lines, and the mean diameter as that rule plus 0.002025, the average of the two ends. Those two numbers make 108.6486, not the printed 107.6486; a rule of 107.64658 would. The mean diameter is the one the rest of the chain uses, and it rounds to the printed 0.2428368 m, so 107.64658 is very likely what was meant.
In water, in air, and the corrections
The report uses 36 of 48 weighings in distilled water, in a vessel packed round with crushed ice, at a mean of 0.3 °C, and 53 weighings in air. From the weight in water it takes away the weight of the air inside the hollow cylinder, then the small buoyancy of the air on the weights; that, subtracted from the weight in air, gives the mass of water displaced: 11.2692387 units. The cylinder’s volume in the cold water, after the brass has shrunk from 17.6 °C and with the submerged part of the small tube that let air in and out, is 11.2796202 cubic decimetres. One divided by the other gives a cubic decimetre of water at 0.3 °C: 0.9990796 of the working unit.
Water is densest near 4 °C, and from weighings at several temperatures Lefèvre-Gineau put the correction at 0.00012757, giving 0.9992072. The pile of Charlemagne, the 50-marc standard of the Paris pound, weighed 12.227944 working units, so one unit is 18,842.088 grains and the kilogram 18,827.15 grains. The table checks twelve of the report’s steps, each from the printed step before it, and arrives at the printed 18,827.15.
| Step | Printed | Recomputed from the printed step before |
|---|---|---|
| Height of the cylinder, lines | 108.061032 | 108.061032 |
| Height, metres | 0.2437672 | 0.2437672 |
| Diameter, metres | 0.2428368 | 0.2428368 |
| External volume at 17.6 °C, dm³ | 11.2900054 | 11.2900055 |
| Volume in the cold water, dm³ | 11.2796202 | 11.2796202 |
| In water, minus the air inside | 0.1968023 | 0.1968023 |
| …and minus the air on the weights | 0.1967668 | 0.1967668 |
| Water displaced | 11.2692387 | 11.2692387 |
| One dm³ of water at 0.3 °C | 0.9990796 | 0.9990796 |
| …at maximum density | 0.9992072 | 0.9992072 |
| The working unit, grains | 18,842.088 | 18,842.088 |
| The kilogram, grains | 18,827.15 | 18,827.15 |
Twelve steps of the report, each recomputed from the printed step before it; the external volume agrees to 0.1 ppm with modern π, the working unit at the three decimals shown (the report prints 18,842.088025; the inputs give 18,842.088253), the rest to their last printed digit. Lengths in lines, 443.296 to the metre; weights in Lefèvre-Gineau’s working unit, close to a kilogram; 4,608 grains to the marc.
Twenty-eight parts in a million
The result was embodied in a platinum cylinder, the kilogram of the Archives, deposited on 22 June 1799; the international prototype of 1889 was matched to it. In 1901 the General Conference on Weights and Measures defined the litre, for precise work, as the volume of one kilogram of pure water at its densest and at standard atmospheric pressure, and measurements of it followed. In 1960 the Conference recorded that this litre and the cubic decimetre differ by about 28 parts in a million, the litre being the larger; in 1964 it abrogated the 1901 definition and let the litre be another name for the cubic decimetre.
So, measured against water, the kilogram fixed in 1799 was about 28 parts in a million too heavy: some 28 milligrams. On this cylinder that is what adding about 2.3 micrometres to both height and diameter would do, as the slider shows. That is an equivalence, not a diagnosis; this page does not know the cause. At the 1907 Conference Charles-Édouard Guillaume called the difference immaterial for volumes found by weighing in commerce and industry. Since 20 May 2019 the kilogram has been defined through the Planck constant.
What the report cannot say
The report gives means, not every weighing; the particular readings in water varied by up to 45 millionths of the unit, and those in air by up to 46. It gives the difference between the two end diameters as 0.0065 millimetres, while its own figures in lines make it 0.0655, and the cone correction it then estimates, 0.0674 cubic millimetres and negligible, fits the larger figure, not the printed one. The page about the meridian that made the metre, on this site, explains where the 443.296 lines came from.
Sources
Every figure on this page is traced in the project’s source notes.
- J. G. Trallès, “Rapport sur l’unité de poids” (30 May 1799), in J. B. J. Delambre (ed.), Base du système métrique décimal, vol. 3, Paris, 1810, pp. 558–580 (here pp. 571–580), facsimile, ETH e-rara.
- CGPM: 1901, Declaration 1; 1960, Resolution 13; 1964, Resolution 6; 1907 proceedings (Guillaume); 2018, Resolution 1.
- The meridian that made the metre, on this site.