Before they would incur so much inconvenience, however, they had recourse to one farther stratagem,—that of trying to set the town on fire. From the height of their mound, they threw down large quantities of fagots, partly into the space between the mound and the newly-built crescent piece of wall,—partly, as far as they could reach, into other parts of the city: pitch and other combustibles were next added, and the whole mass set on fire. The conflagration was tremendous, such as had never been before seen: a large portion of the town became unapproachable, and the whole of it narrowly escaped destruction. Nothing could have preserved it, had the wind been rather more favorable: there was indeed a farther story, of a most opportune thunder-storm coming to extinguish the flames, which Thucydidês does not seem to credit.[331] In spite of much partial damage, the town remained still defensible, and the spirit of the inhabitants unsubdued. [331] Thucyd. ii, 77. There now remained no other resource except to build a wall of circumvallation round Platæa, and trust to the slow process of famine. The task was distributed in suitable fractions among the various confederate cities, and completed about the middle of September, a little before the autumnal equinox.[332] Two distinct walls were constructed, with sixteen feet of intermediate space all covered in, so as to look like one very thick wall: there were, moreover, two ditches, out of which the bricks for the wall had been taken,—one on the inside towards Platæa, and the other on the outside against any foreign relieving force. The interior covered space between the walls was intended to serve as permanent quarters for the troops left on guard, consisting half of Bœotians and half of Peloponnesians.[333] [332] Thucyd. ii, 78. καὶ ἐπειδὴ πᾶν ἐξείργαστο περὶ Ἀρκτούρου ἐπιτολάς, etc. at the period of the year when the star Arcturus rises immediately before sunrise,—that is, sometime between the 12th and 17th of September: see Göller’s note on the passage. Thucydidês does not often give any fixed marks to discriminate the various periods of the year, as we find it here done. The Greek months were all lunar months, or nominally so: the names of months, as well as the practice of intercalation to rectify the calendar, varied from city to city; so that if Thucydidês had specified the day of the Attic month Boêdromion (instead of specifying the rising of Arcturus) on which this work was finished, many of his readers would not have distinctly understood him. Hippokratês also, in indications of time for medical purposes, employs the appearance of Arcturus and other stars.