Altaic is the standard example in linguistics of a proposed family that did not hold. In its usual form it joins Turkic, Mongolic and Tungusic; in wider versions it adds Korean and Japonic and is sometimes called Transeurasian. The proposal is old, dating from the nineteenth century, and it was mainstream for much of the twentieth. The evidence for it is real and consists of shared vocabulary, shared vowel harmony and broadly similar grammatical structure. The case against it is that all three of those are exactly what centuries of contact between neighbouring steppe languages would produce, and that the specific kind of evidence a family requires, basic vocabulary showing regular sound correspondences, and shared irregular morphology, has not been assembled to the standard the comparative method demands. The strongest form of the argument against it is that the shared words cluster in the semantic fields associated with borrowing, herding, warfare and material culture, rather than in the core vocabulary that resists it.
Most Turkologists, Mongolists and Tungusologists now work without the hypothesis, and the atlas carries it as a proposal with its classification marked debated. The name has an additional problem. Altaic was coined from the Altai mountains on an assumption about where the ancestral speakers lived, so the label asserts a homeland for a family that has not been shown to exist, which is one reason recent work defending the grouping prefers the neutral term Transeurasian.