A verb for incomplete knowledge
To guess is to reach a conclusion without enough evidence to be sure of it. English uses the verb for a stab at a riddle, a rough count, and the everyday hedge I guess, which often means little more than I suppose. The matching noun names the attempt itself: a wild guess, an educated guess, guesswork.
That range looks simple until the history is unpacked. The origin of the word guess is not a native Old English survival. It is a Germanic loan that English picked up in the Middle Ages, and the trail still has a fork in it.
Middle English gessen
The verb appears around 1300 as gessen. Early uses cover inferring from what one can see, judging, deciding, and estimating size or number. One of the earliest cited English examples is from about 1320, in the romance King Horn. By the fourteenth century the word is common and geographically wide.
There is no matching Romance verb. The nearest relatives sit in North and West Germanic. Middle Danish had gitse and getze; later Danish has gisse, Swedish and Norwegian gissa, Icelandic giska. On the continent, Middle Dutch had gessen and Middle Low German gissen, with Dutch gissen still in ordinary use.
Most dictionaries therefore treat English gessen as a borrowing, probably from early Scandinavian, perhaps from Low German, or from one of those streams under the influence of the other. A phonological argument favors Old Norse: a native Old English cognate of this type would be expected to palatalize the initial /g/, which did not happen. That makes a Scandinavian source attractive. The same comparative work also records the counter-suggestion that Scandinavian itself, and conceivably English, might have taken the verb from Middle Low German. English almost certainly borrowed the word. Which North Sea neighbor supplied it is not settled beyond dispute.
The noun is formed in English by conversion from the verb, though Middle English already had gesse. Around 1300 it could mean a doubtful supposition or mere guesswork. By the mid-fifteenth century it could also mean a considered opinion.
From get to estimate
Under the surface, guess is a relative of get. The Germanic forms are generally derived from Proto-Germanic *getaną, to get, itself from a reconstructed Indo-European root meaning to take or seize. Some dictionaries reconstruct an intermediate *gitisōną for the guessing verb; the exact oldest shape is still a matter of reconstruction.
The extra consonants are the useful clue. Danish gitse shows that the English -ss once reflected -ts, a remnant of a suffix attached to the get root. Icelandic giska is sometimes analyzed as get plus a reflexive element -sk, which would give a sense like to get something for oneself; that analysis belongs to the Icelandic form and should not be forced onto English. An older reconstruction of the early meaning runs along the lines of to try to get, or to be ready to get. A common sketch of the prehistoric shift is get, then take aim at, then estimate: you try to obtain the thing you are after, then you aim at it, then you size it up without having it in hand. In that light the original idea was to get what you are looking for.
English get itself belongs to the same family. Old English used the related verb mostly in compounds, which survive as beget and forget. The isolated verb guess, by contrast, is the newcomer. Cognates keep the family visible: Dutch gissen, Low German gissen, Saterland Frisian gisje, Danish gisse and gisne, Norwegian gissa and gjette, Swedish gissa. Old Norse geta already covered both get and guess.
How the sense narrowed and spread
Middle English gessen was broader than modern schoolbook conjecture. It could mean perceive, find out, form an opinion, judge, and estimate. Hitting on the right answer is a later specialization, recorded from the 1540s. That is the sense behind a guessing game, attested from the 1650s, and behind keep someone guessing, American English from the 1890s. The shrug your guess is as good as mine is later still, from the start of the twentieth century.
An older sense, now obsolete, was to reproduce something from memory, as when a speaker is asked to recover words as nearly as they can. In the other direction, guess once commonly meant simply to suppose or to believe. That milder use is still the force of American I guess, the first-person filler that answers a question with I guess so rather than with a calculated estimate. Commentators in the nineteenth century treated the habit as especially New England, but the suppose sense is older than that regional label. It was already available in early modern English, including in the meaning to think, and it survived in British speech into the eighteenth century. What changed is not the existence of the hedge, but how far it crowded out the older, sharper uses. To guess a riddle, once ordinary, has largely given way to solve.
The spelling with gu- is late sixteenth century. English ge- and gi- are unreliable guides to pronunciation, as get and gesture show, and the extra u marks a hard /g/. The spelling is sometimes credited to William Caxton and his years as a printer in Bruges, where Dutch-looking forms could influence English type. That attribution is traditional rather than proven.
The second s
The concrete leftover is the second s. It is not decoration. It is what remains of the old -ts suffix that turned the get root into a verb of aiming and estimating. English speakers who say guess are still using a word built to mean getting at something they do not yet have.