Before the Enlightenment, clouds were described purely phenomenologically. Old woodcuts show God’s punishing hand emerging from storm clouds to strike sinners, while printers sold single-sheet broadsheets depicting battleships, fighting knights, raging horses, dragons and angels imagined into cloud shapes. Sailors called cirrus clouds wind trees, cat’s tails or mare’s tails; in England a mackerel sky meant fleecy clouds, and mammatus formations were known as smallpox clouds in the Orkneys. For a long time clouds were even understood as rigid bodies, and people believed that firing cannons could alter the path of thunderclouds. It was the naturalist Heinrich Wilhelm Dove, in the nineteenth century, who reframed the problem entirely, arguing that clouds are dynamic processes constantly forming and disappearing, and should be considered as flow rather than as solid objects.

The scientific study of clouds arguably began with Otto von Guericke, the inventor of the vacuum pump and mayor of Magdeburg. In his work Experimenta Nova, completed in 1663 and printed in 1672, he described the first experimental production of a cloud: a large evacuated vessel was connected to a smaller vessel containing moist air at atmospheric or higher pressure. When the connecting tap was opened, the air expanding into the vacuum cooled adiabatically, and mist droplets became visible, slowly sinking down. Guericke assumed, without further reasoning, that these droplets were hollow vesicles. That single assumption would haunt atmospheric science for nearly two hundred years.

The vesicle hypothesis spread rapidly. Edmond Halley in London used it in 1690, imagining that when water boils, vapour bubbles leave the surface as hollow vesicles filled with a light gas, becoming specifically lighter than air and therefore able to float. A young Gottfried Wilhelm Leibniz studied Guericke’s book intensively, corresponded with its author, and in 1710 wrote his own treatise on the elevation of vapours, likewise assuming hollow vesicles lighter than air. The philosopher Christian Wolff, observing that droplets appeared white when crossing a beam of light, reasoned by analogy with white foam on brown beer that floating droplets must also be hollow. In 1743, Christian Gottlieb Kratzenstein won a prize from the Académie des Sciences de Bordeaux by arguing that a rainbow can never be observed in a cloud, so cloud droplets cannot be compact but must be hollow, and that buoyancy from surrounding air particles should carry vesicles upward until falling air pressure brought them to a standstill.

The dogma proved remarkably durable. Leading scientists including Euler, Saussure, Humboldt, Volta, Kämtz and Clausius held on to it, and even the philosopher Arthur Schopenhauer regarded cloud vesicles as real. Doubts accumulated slowly: Ludwig Wilhelm Gilbert noted in 1805 that only water vapour could be found inside bubbles, meaning they would have to sink rather than stay suspended; Augustus Waller failed to detect vesicles under the microscope in 1847; and Kober showed through microscopic examinations in 1871 that vesicles did not exist in the atmosphere, though his dissertation attracted little attention. Only in 1885 did Richard Assmann examine real cloud droplets on top of the Brocken mountain in the Harz with a microscope and find no evidence of hollow vesicles, finally refuting the hypothesis in the Meteorologische Zeitschrift, one of the leading journals of the day. Even then, many articles continued to uncritically adhere to the old idea.

Meanwhile, the classification of clouds was taking shape. The mathematician Friedrich Meister in Göttingen made an early attempt in 1780, distinguishing bulbous, convex shapes from holey, concave ones. Jean Baptiste de Lamarck proposed five cloud types across three atmospheric levels in 1802, a vertical division still used today, though his French names never caught on. The decisive breakthrough came in 1803, when Luke Howard, an English businessman who manufactured pharmaceutical chemicals and observed clouds as a hobby, introduced a systematic scheme with Latin terminology. Because Latin remained the scholarly language of Europe, Howard’s terms had international appeal and eliminated local trivial names. He defined three basic types and supplemented them with mixed forms, distinguishing seven cloud types in total. His scheme laid the cornerstone for today’s internationally recognized cloud classification, though he himself overestimated the role of electricity at the cirrus level, reasoning from the resemblance of cirrus forms to Lichtenberg’s electrostatic dust figures.

Physical understanding advanced through thermodynamics. James Pollard Espy recognized that adiabatic processes occur in the atmosphere due to the conservation of energy, and that the release of latent heat during condensation drives thunderstorm formation and convection in low-pressure areas. Using an instrument called a nephelescope, similar in principle to Guericke’s apparatus, he demonstrated that cumulus clouds gain additional buoyancy through latent heat release. Hermann Hertz developed the first adiabatic chart in 1884, allowing rapid visual assessment of dry and moist-adiabatic states, and more than twenty different adiabatic diagrams were eventually created, becoming indispensable for the rapidly growing aviation industry after World War I. Heinrich von Bezold improved the theory of atmospheric thermodynamics and warned against rushing to create artificial cloud classifications on purely descriptive grounds, a caution aimed at observers who had distinguished as many as 228 different cloud types.

The exploration of the third dimension transformed the field. Manned balloon flights from 1880 onwards provided reliable vertical profiles of temperature and humidity within and above clouds, though early instruments suffered serious radiation biases and acceleration errors until Assmann’s aspirated psychrometer of 1892. Balloonists observed the coexistence of supercooled water drops and snow crystals below freezing level, a finding Alfred Wegener used to explain how snow crystals grow at the expense of supercooled droplets. In 1894 the International Meteorological Commission established the International Commission for Cloud Research, and during the International Cloud Year of 1896/97, scientists from eleven, eventually twenty-seven, nations measured cloud heights and motions using paired theodolites, photogrammetry and specially developed cloud cameras. Hildebrandsson’s evaluation of upper-level winds revealed divergent flow over lows and convergent flow over highs, connecting cloud movement to vertical motion in pressure systems.

The Bergen cyclone model developed by Jakob Bjerknes in 1919 gave observers a powerful framework: with an approaching warm front, stratiform cirrus appears first, thickening to altostratus and finally nimbostratus with rain, while behind a cold front, high-reaching cumulus clouds develop. Cloud shapes could now be read as diagnostics of atmospheric stability, and certain forms became reliable forecast tools, such as altocumulus castellanus, which signals afternoon thunderstorms. Depicting clouds accurately remained difficult, however, since early photography could barely distinguish clouds from blue sky; yellow filters and polarization tricks helped, but painted images and watercolours, including a remarkable collection of 400 cloud paintings by the pharmacist Ernst Mylius, were used in atlases well into the twentieth century. After the founding of the World Meteorological Organization in 1951, the first modern International Cloud Atlas appeared in 1956, standardizing ten cloud genera and fifteen species. The study also recalls discoveries beyond the troposphere, from polar stratospheric clouds at 21 to 27 kilometres to noctilucent clouds near the mesopause at roughly 80 to 89 kilometres, and notes that even today, satellite missions such as the European Space Agency’s recently selected WIVERN explorer are revisiting the oldest questions in the sky.

Subject of Research: The 300-year history of cloud observation, classification and the physics of cloud formation

Article Title: A 300-year history of understanding and classifying clouds, from a German language perspective

Article References: Winkler, P. (2026). A 300-year history of understanding and classifying clouds, from a German language perspective. History of Geo- and Space Sciences, 17(1), 37-59. https://doi.org/10.5194/hgss-17-37-2026

Image Credits: AI Generated

DOI: 10.5194/hgss-17-37-2026

Keywords: clouds, cloud classification, history of meteorology, Otto von Guericke, Luke Howard, vesicle hypothesis, balloon ascents, International Cloud Atlas, World Meteorological Organization, adiabatic processes, noctilucent clouds, Bergen cyclone model