Why is the thing dual to a "meridian" called a "longitude"? A pair of distinguished generators of the fundamental group $\pi_1(\partial(S^3 \setminus K))$ of the boundary torus of a knot complement are usually called the "meridian" and "longitude". However, this terminology has always seemed a bit odd to me: in geography a meridian is a line of longitude, so shouldn't the curve it intersects be a "latitude"? Does anyone know the origin of this language?
 A: If a vector points northward, then it generates a meridian of longitude. Following that curve keeps one at the same longitude. A vector pointing eastward points in the direction of a curve whose points are at different longitudes. You mark the points on that curve with their longitudes.
Suppose a person's income were a function of the person's height and you want to graph that function. Label the horizontal axis "height" and the vertical axis "income". The axis labeled "height" is a line along which the height varies. It is not a line of constant height. Likewise the thing labeled "longitude" is a curve along which the longitude varies, not a curve of constant longitude.
A: There is a fundamental asymmetry between latitude and longitude on a sphere, whereas on a torus, there is a symmetry between the two generators.  This symmetry could motivate the use of nearly synonymous words.
The terminology may originate with the paper On the homology invariants of knots by H. Seifert (Quart. J. Math. Oxford (2) 1 (1950), 23–32).  The torus $T$ of interest to Seifert is the boundary of a closed tubular neighborhood $V$ of a knot in $\mathbb{R}^3$.  Seifert says that a Jordan curve on $T$ that bounds on $V$ (respectively, on $\mathbb{R}^3 - V + T$) but not on $T$ is called a meridian (respectively, a longitudinal circuit). Seifert phrases these definitions in a way that highlights the symmetry, which makes the use of near synonyms seem natural.
