The claim begins with four words. In the introduction to his Handbook of American Indian Languages (1911), Franz Boas gave an Inuit example: aput "snow on the ground", qana "falling snow", piqsirpoq "drifting snow" and qimuqsuq "a snowdrift". His point was about word roots, not perception. English uses unrelated words for forms of water (lake, river, brook, rain, dew, foam) where it could have derived them from one stem, and Inuit does the same with snow. Boas had wintered among the Inuit of Baffin Island in 1883–84 and knew the vocabulary first-hand.
From four to one hundred
In April 1940 Benjamin Lee Whorf published "Science and Linguistics" in MIT's Technology Review (vol. 42, no. 6). He listed falling snow, snow on the ground, snow packed hard like ice, slushy snow and wind-driven flying snow as things an Eskimo language names separately, and claimed that English uses one word for all of them. As later critics counted it, that turned Boas's four roots into at least seven, and it passed over English slush, sleet and blizzard. Later popular books kept raising the number. On 9 February 1984 an unsigned New York Times editorial put it at one hundred.
The anthropologist Laura Martin traced this history in "'Eskimo Words for Snow': A Case Study in the Genesis and Decay of an Anthropological Example" (American Anthropologist 88:2, 1986). She showed that each retelling cited the previous one instead of a dictionary. In 1989 Geoffrey K. Pullum brought her findings to a wider audience with "The Great Eskimo Vocabulary Hoax" (Natural Language and Linguistic Theory 7:2), later the title essay of a 1991 book. Pullum argued that Eskimo languages and English have roughly the same number of snow roots.
What the dictionaries hold
Part of the confusion is about what counts as a word. Inuit and Yupik languages are polysynthetic: suffixes stack on a root to make words that English renders as whole sentences. On that definition the number of snow "words" has no upper limit, and the same holds for any other noun. Counting roots gives a much smaller figure. In 1991 Anthony Woodbury went through Steven Jacobson's 1984 Central Alaskan Yup'ik dictionary and found 15 snow-related meanings expressed by about 33 forms. They include qanuk "snowflake", qanikcaq "snow on the ground", muruaneq "soft, deep snow", nutaryuk "fresh snow" and qanisqineq "snow floating on water". His English comparison list for the same purpose ran to 22 terms.
The map shows three snow roots that the Comparative Eskimo Dictionary (2nd ed., 2010) reconstructs for Proto-Eskimo: *apun "snow on the ground" (Greenlandic and Inuktitut aput, Iñupiaq apun), *qaniɣ "falling snow" (behind Yup'ik qanikcaq) and *aniɣu "fallen snow" (Central Siberian Yupik анигу). The basic snow word differs between neighbouring languages because each picked a different root out of that shared set.
Where the real richness is
Debunking the snow figure did not mean these languages lacked specialised vocabulary. The anthropologist Igor Krupnik of the Smithsonian helped lead the SIKU (Sea Ice Knowledge and Use) project during the International Polar Year of 2007–2008, which recorded ice and snow terms with hunters in about ten Inuit and Yupik dialects. As David Robson reported in New Scientist in 2012, the Nunavik dialect of Inuktitut in Quebec has at least 53 snow terms. They include matsaaruti, wet snow used to ice sled runners, and pukak, crystalline powder snow that looks like salt. Central Siberian Yupik has about 40. The Iñupiaq of Wales, Alaska, has about 70 terms for ice, including utuqaq for ice that lasts year after year and auniq for ice full of holes.
On St. Lawrence Island, the Yupik hunter Conrad Oozeva wrote down nearly 100 sea-ice terms in 1986 for school language classes. The Smithsonian published them in 2004 as an illustrated dictionary in Watching Ice and Weather Our Way. In the SIKU volume (2010), Krupnik and Ludger Müller-Wille went back to Boas's Baffin Island diaries and field notes. They argued that the hoax debate had misread Boas: he had used four terms to make a point about grammar, and never claimed to be listing the whole vocabulary. The large inventories are mostly about ice, and they sort it by age, thickness and whether it will hold a person or a boat.