The festival took place at the second or third full moon after the summer solstice, in the months of Apollonios and Parthenios respectively.[284] Its date was fixed by a cycle of eight years or ninety-nine-months, the divergence between the year of twelve lunar months and the solar year being rectified by the insertion of three intercalary months, one in the first four years, two in the second. Thus it fell alternately after forty-nine or fifty lunar months. The fourteenth day of the month seems to have been reckoned as the day of the full moon, though the actual full moon varied from the 14th to 15th. This day must, from the earliest time, have been the central day of the festival.[285] The Greek day was reckoned from sunset to sunset, and as Greek custom demanded that sacrifice to the Olympian gods should be offered in the morning, before mid-day,[286] it follows that the great sacrifice to Zeus was offered on the morning after the full moon. The festival lasted five days. According to Herodotus, a historian of the fifth century, the five days’ festival was ordained by Heracles.[287] Certainly it lasted five days in Pindar’s time.[288] Scholiasts of various dates, while affirming that it lasted five days, state that it began on the 10th or 11th and lasted till the 15th or 16th.[289] The discrepancy may be due to the variation in the date of the full moon already noticed, more probably to the addition to the festival of one or more preliminary days necessitated in later times by the multiplication of competitions and religious ceremonies. To these days the preliminary business of the festival may have been transferred, but they were not reckoned as part of the actual festival. The seventh ode of Bacchylides, written in honour of Laches of Ceos, who won the boys’ foot-race in 452 B.C., proves beyond doubt that in this year the festival ended on the sixteenth day. If then the festival lasted five days, the fourteenth, the day of the full moon, was the central day of the whole festival. The recognition of the importance of this fact is due to Ludwig Weniger, whose conclusions I have in the main adopted in the following pages.