Plutarch, Caesar 59:
59. 1. The adjustment of the calendar, however, and the correction of the irregularity in the computation of time, were not only studied scientifically by him, but also brought to completion, and proved to be of the highest utility.
2. For not only in very ancient times was the relation of the lunar to the solar year in great confusion among the Romans, so that the sacrificial feasts and festivals, diverging gradually, at last fell in opposite seasons of the year, 3. but also at this time people generally had no way of computing the actual solar year; the priests alone knew the proper time, and would suddenly and to everybody’s surprise insert the intercalary month called Mercedonius.
4. Numa the king is said to have been the first to intercalate this month, thus devising a slight and short-lived remedy for the error in regard to the sidereal and solar cycles, as I have told in his Life.
5. But Caesar laid the problem before the best philosophers and mathematicians, and out of the methods of correction which were already at hand compounded one of his own which was more accurate than any. This the Romans use down to the present time, and are thought to be less in error than other peoples as regards the inequality between the lunar and solar years.
6. However, even this furnished occasion for blame to those who envied Caesar and disliked his power. At any rate, Cicero the orator, we are told, when some one remarked that Lyra would rise on the morrow, said: “Yes, by decree,” implying that men were compelled to accept even this dispensation.
Macrobius, Saturnalia, bk 1, ch. 14:
2. Subsequently, however, since there was thus inconsistency in the marking of the times and seasons but all was still vague and uncertain, Gaius Caesar introduced a clearly defined arrangement of the calendar, with the help of a clerk named Marcus Flavius, who provided the dictator with a list of the several days so arranged that their order could be easily found and, that order once found, the position of each day would remain constant.
3. Caesar therefore began the new arrangement of the calendar by using up all the days which could still have caused confusion, with the result that the last of the years of uncertainty was prolonged to one of four hundred and forty-three days. Then, copying the Egyptians – the only people who fully understood the principles of astronomy – he endeavored to arrange the year to conform to the duration of the course of the sun, which it takes three hundred and sixty-five days and a quarter to complete.
4. For just as the lunar cycle is the month, since the moon takes rather less than a month to make a circuit of the zodiac, so the solar cycle must be reckoned by the number of days which the sun takes to turn again to that sign of the zodiac from which it began its course. That is why the common year is styled the “turning” year and is held to be the “great” year (since the lunar cycle is thought of as the “short” year), 5. and Vergil has combined these two descriptions of the solar year in the line:
Meanwhile the sun completes the turning of the great year. [Aeneid 3. 284]
It is for this reason that Ateius Capito too thinks that the word “year” (annus) is to be explained as a circuit of time; namely, because of old an used to stand for “around,” as, for example, where Cato in his Origins writes: “Let the plough be driven around the boundary,” using an instead of circum; or when we say ambire
6. Julius Caesar therefore added ten days to the old arrangement of the calendar, in order that the year might consist of the three hundred and sixty-five days which the sun takes to pass through the zodiac; and, to allow for the remaining quarter of a day, he ordained that the priest in charge of the months and days should insert one day every fourth year in that month, and in that part of it, in which of old an intercalary month used to be inserted, that is to say, immediately before the last five days of February. This intercalary day he ordered to be called bissextus [as doubling the sixth day before the Kalends of March].
7. The arrangement to distribute the ten additional days to which I have referred was as follows: January, Sextilis, and December received two days each, and April, June, September, and November one each. No addition was made to the month of February, lest changes in connection with the worship of the gods below might result; and March, May, Quintilis, and October remained as they had been of old, because they already had the full complement of thirty-one days apiece.
8. And, since Caesar made no change in these four months, they also have the Nones on the seventh day, as laid down by Numa. But in January, Sextilis, and December, the months to which Caesar added two days apiece, although after his reforms each for the first time had thirty-one days, nevertheless the Nones come on the fifth day and the Kalends that follow return on the nineteenth day after the Ides, because Caesar would not insert the additional days before either the Nones or the Ides for fear that an unprecedented postponement by two days (which would be the result of such change) might interfere with religious ceremonies appointed to be held on a day fixed in relation to the Nones or Ides.
9. Nor yet would he insert the additional days immediately after the Ides for fear of disturbing appointed rest days, but a place was not made for them in any month until the celebration of the rest days held in that
month had been completed. Thus in January the allotted days to which we refer were the fourth and third days before the Kalends of February; in April, the sixth day before the Kalends of May; in June, the third day before the Kalends of July; in August, the fourth and third day before the Kalends of September; in September, the third day before the Kalends of October; in November, the third day before the Kalends of December; and in December, the fourth and third days before the Kalends of January.
10. Consequently, although, before this reform, in all the months to which days were added the Kalends of the following months returned on the seventeenth day after the Ides; afterward, as the result of the additions, the Kalends returned on the nineteenth day after the Ides in the months which received two days and on the eighteenth in the months which received one.
11. In each month, however, rest days kept their appointed places. For example, if the third day after the Ides was generally observed as a festival or a rest day and used formerly to be known as the sixteenth day before the following Kalends, even after the number of days in the month had been increased, the religious observance remained unchanged and the ceremony was still held on the third day after the Ides, although (in consequence of an increase in the number of days in the month) the day was no longer the sixteenth day before the following Kalends but the seventeenth, if one day had been added to the month, and the eighteenth, if two days had been added.
12. That is why Caesar inserted the new days, in each case, toward the end of the month, at a time when all the rest days in the month were found to be over. Moreover, he caused these additional days to be marked in the calendar as fasti, so as to make more time available for legal business; and he not only arranged that all these days should be such days of legal business but also that none should be a day on which an assembly might be held, his intention being that this increase in the number of the days should not add to a magistrate’s power to exercise undue influence.
13. Caesar’s regulation of the civil year to accord with this revised measurement was proclaimed publicly by edict, and the arrangement might have continued to stand had not the correction itself of the calendar led the priests to introduce a new error of their own; for they proceeded to insert the intercalary day, which represented the four quarter-days, at the beginning of each fourth year instead of at its end, although the intercalation ought to have been made at the end of each fourth year and before the beginning of the fifth.
14. This error continued for thirty-six years, by which time twelve intercalary days had been inserted instead of the number actually due, namely, nine. But, when this error was at length recognized, it too was corrected, by an order of Augustus that twelve years should be allowed to pass without an intercalary day, since a sequence of twelve such years would account for those three days too many which, in the course of the thirty-six years, had been introduced by the premature action of the priests.
15. After that, one intercalary day, as ordered by Caesar, was to be inserted at the beginning of every fifth year, and the whole of this arrangement of the calendar was to be engraved on a bronze tablet, to ensure that it should always be observed.
Pliny the Elder, Natural History, bk. 18:
56. There follows the question postponed to this place, a question that needs very careful consideration – that of the proper date for sowing the crops; it is in a large degree connected with astronomy, and consequently we will begin by setting out the views of all authors in regard to it. …
57. First of all it is almost impossible to explain the system of the actual days of the year and that of the movement of the sun, because to the 365 days an intercalary year adds a quarter of a day and of a night, and consequently definite periods of the stars cannot be stated. In addition to this there is the admitted obscurity of the facts, as sometimes the specification of the seasons runs in advance, and by a considerable number of days … , whereas at other times it comesbehind … and in general the influence of the heavens falls down to the earth in one place more quickly and in another place more slowly; this is the cause of the remark we commonly hear on the return of fine weather, that a constellation has been completed. Moreover although all these things depend on stars that are stationary and fixed in the sky, there intervene movements of stars and hailstorms and rain, these also having no inconsiderable effect, as we have shown, and they disturb the regularity of the expectation that has been conceived. …
Additional difficulty has also been caused by authors through their observations having been taken in different regions, and because in the next place they actually publish different results of observations made in the same regions. But there were three main schools, the Chaldaean, the Egyptian and the Greek; and to these a fourth system was added in our own country by Caesar during his dictatorship, who with the assistance of the learned astronomer Sosigenes brought the separate years back into conformity with the course of the sun – and this theory itself was afterwards corrected (when an error a had been found), so as to dispense with an intercalary day for a period of twelve successive years, for the reason that the year which had previously been getting in advance of the constellations had begun to lag behind in relation to them.
Both Sosigenes himself in his three treatises – though more careful in research than the other writers he nevertheless did not hesitate to introduce an element of doubt by correcting his own statements – and also other authors whose names we prefixed to this volume have published these theories, although it is seldom that the opinions of any two of them agree.
… the morning setting of the Pleiads is given by Hesiod – for there is extant an astronomical work that bears his name also – as taking place at the close of the autumnal eqninox, whereas Thales puts it on the 5th day after the equinox, Anaximander on the 30th, Euctemon on the 44th, and Eudoxus on the 48th. We follow the observation of Caesar specially: this will be the formula for Italy; but we will also state the views of others, …
70. From midwinter till the west wind blows the important stars that mark the dates, according to Caesar’s observations, are – the Dogstar setting at dawn on December 30, the day on which the Eagle is reported to set in the evening for Attica and the neighbouring regions; on January 4 according to Caesar’s observations the Dolphin rises at dawn and the next day the Lyre, the Arrow setting in the evening on the same day for Egypt …
75. Between the period of west wind and the spring equinox, February 16 for Caesar marks three days of changeable weather, as also does February 22 by the appearance of the swallow and on the next day the rising of Arcturus in the evening, and the same on March 5 – Caesar noticed that this bad weather took place at the rising of the Crab, but the majority of the authorities put it at the setting of the Vintager – on March 8 at the rising of the northern part of the Fish, and on the next day at the rising of Orion; in Attica it is noticed that the constellation Kite appears. Caesar also noted March 15 – the day that was fatal to him – as marked by the setting of the Scorpion, but stated that on March 18 the Kite becomes visible in Italy and on March 21 the Horse sets in the morning. …
I’ve added a little more from Pliny than strictly necessary, as it indicates that Caesar’s calendar was not merely what we think of as the Julian calendar, but comprised a whole series of astronomical notations, for the purpose of crop management. No doubt Sosigenes compiled these, but it is interesting to see them.
-  Translated by Percival Vaughan Davies, Columbia University Press, 1969. ↩