The meaning of the first part is plain; an epigram may be complete within the limits of a single couplet. But do "the three" mean three lines or three couplets? "Exceeding three" would, in the one case, mean an epigram of four lines, in the other of eight. As there cannot properly be an epigram of three lines, it would seem rather to mean the latter. Even so the statement is an exaggeration; many of the best epigrams are in six and eight lines. But it is true that the epigram may "have its nature", in the phrase of Aristotle,[6] in a single couplet; and we shall generally find that in those of eight lines, as always without exception in those of more than eight, there is either some repetition of idea not necessary to the full expression of the thought, or some redundance of epithet or detail too florid for the best taste, or, as in most of the Byzantine epigrams, a natural verbosity which affects the style throughout and weakens the force and directness of the epigram. The notorious difficulty of giving any satisfactory definition of poetry is almost equalled by the difficulty of defining with precision any one of its kinds; and the epigram in Greek, while it always remained conditioned by being in its essence and origin an inscriptional poem, took in the later periods so wide a range of subject and treatment that it can perhaps only be limited by certain abstract conventions of length and metre. Sometimes it becomes in all but metrical form a lyric; sometimes it hardly rises beyond the versified statement of a fact or an idea; sometimes it is barely distinguishable from a snatch of pastoral. The shorter pieces of the elegiac poets might very often well be classed as epigrams but for the uncertainty, due to the form in which their text has come down to us, whether they are not in all cases, as they undoubtedly are in some, portions of longer poems. Many couplets and quatrains of Theognis fall under this head; and an excellent instance on a larger scale is the fragment of fourteen lines by Simonides of Amorgos,[7] which is the exact type on which many of the later epigrams of life are moulded. In such cases /respice auctoris animum/ is a safe rule; what was not written as an epigram is not an epigram. Yet it has seemed worth while to illustrate this rule by its exceptions; and there will be found in this collection fragments of Mimnermus and Theognis[8] which in everything but the actual circumstance of their origin satisfy any requirement which can be made. In the Palatine Anthology itself, indeed, there are a few instances[9] where this very thing is done. As a rule, however, these short passages belong to the class of {gromai} or moral sentences, which, even when expressed in elegiac verse, is sufficiently distinct from the true epigram. One instance will suffice. In the Anthology there occurs this couplet:[10] {Pan to peritton axaipon epei logos esti palaios os xai tou melitos to pleon esti khole}