Address to Pandrosion: Theorem, Problem, and the Threefold Hierarchy of Loci¶
Synagoge Book III, Chs. 1 & 20–22 (Hultsch I, pp. 30–32, 54–56) · Original English Translation from Hultsch’s Greek
Those of the ancients who wished to examine geometric inquiries with precision, most noble Pandrosion, judged it necessary to distinguish a problem (πρόβλημα) from a theorem (θεώρημα). He who proposes a theorem—whatever be the geometric matter set out and whatever attribute he claims follows from those hypotheses—invites us to behold how that consequence holds of necessity; whereas he who sets forth a problem bids us construct something not yet brought into view, and therefore stands under an obligation to determine by a diorismos when, how, and in how many ways the construction is possible. If anyone announces a problem whose construction cannot be carried out by the means he assumes, he errs doubly: first in mistaking the nature of the locus to which the inquiry belongs, and second in offering a defective demonstration to learners.
Now some of those who came to us purporting to have studied geometry with you set out an attempted plane construction for finding two mean proportionals in continued proportion between two given unequal straight lines—the classic Delian problem of duplicating the cube, since Hippocrates of Chios and Euclid (Elements XI.33) showed that if $a : x = x : y = y : b$, then $a : b = a^3 : x^3$. When we tested their construction rigorously (Chs. 2–19, Hultsch I pp. 32–54), their assumed intersection failed to meet the required proportional condition, because the problem is by nature solid rather than plane.
For we affirm (Chs. 20–21, Hultsch I pp. 54–56) that there are three genera of geometric problems (ἐπίπεδα, στερεά, γραμμικά):
Plane problems (ἐπίπεδα) are those capable of being solved exclusively by means of the straight line and the circumference of the circle; they are justly called plane because the lines through which they are found have their generation in a plane.
Solid problems (στερεά) are those whose solution requires the employment of one or more sections of the cone—parabola, hyperbola, or ellipse—for their construction demands surfaces of solid figures, namely conical surfaces.
Linear problems (γραμμικά) comprise a third genus still remaining beyond these: for their construction requires lines other than those just spoken of, possessing a more varied and forced generation (ποικιλωτέραν καὶ βεβιασμένην τὴν γένεσιν) arising from irregular surfaces and interwoven motions—such as the spirals (ἕλικες) of Archimedes, the quadratrices (τετραγωνίζουσαι) of Dinostratus, the conchoids (κογχοειδεῖς) of Nicomedes, and the cissoids (κισσοειδεῖς).
Geometers consider it no small methodological fault (ἁμάρτημα οὐ μικρόν) whenever anyone solves a plane problem by conics or higher linear curves, or conversely attempts to force a solid problem—such as the duplication of the cube or the trisection of an arbitrary rectilinear angle—into straight lines and circles alone. Because conic sections are delicate to draw with instruments in the plane, the ancients also devised exact mechanical and neusis constructions for two mean proportionals (Chs. 22–27, Props. 1–5, Hultsch I pp. 56–68): the sliding three-plate mesolabe of Eratosthenes, the conchoidal verging ruler of Nicomedes, the rectangle-and-pivot rule of Hero of Alexandria, and our own semicircular construction equivalent in geometric effect to Hero’s.