This last allusion is clearly to a partial eclipse of the sun, which must have been observed this year, U. C. 537 Varr. B. C. 217. The ingress of the consular year indeed is mentioned by Livy before he notices this fact: Per idem tempus Cn. Servilius consul Romæ Idibus Martiis magistratum iniit. But the Ides of March this year fell on April 5; and if this eclipse happened in the February preceding, it is very possible that the above report of its having been observed, somewhere at a distance from Rome, and in the winter time, might not reach Rome until after April 5.

There was only one solar eclipse, according to Pingré, B. C. 217; viz. on February 11: which he has determined to 2.30 P. M. for the meridian of Paris. We have calculated this ecliptic conjunction ourselves; and determined it to February 11, at 13 h. 58 m. from midnight for the meridian of Greenwich. The sun's mean distance from the ascending node at the time was 5 s. 23° 22′; so that there must have been a considerable eclipse on this occasion somewhere or other. And this being assumed accordingly as the eclipse intended by Livy in this instance, its date in the regular calendar of the time, though not specified by him, must have been xiii Kalendas Februarias U. C. 537 Varr. January 18 Roman.

And this too being supposed to have been the actual Roman date of the phenomenon in question, xiii Kalendas Februarias; it is very observable that the permanent attachment of the Saturnalia to one day of the month, (the same which is ever after known as their stated date,) the xiv Kalendas Januarias, December 17 Roman, appears to have ultimately arisen out of it. Postremo, observes Livy, Decembri jam mense ad ædem Saturni Romæ inmolatum est, lectisterniumque imperatum. . . ac per urbem Saturnalia diem ac noctem clamatum; populusque eum diem festum habere ac servare in perpetuum jussus. There can be no doubt that the December here alluded to was that of this same consular year; and that the procuration of the prodigies in question including that of the eclipse for some reason or other lasted until then. And thus much is implied even in the mode in which Livy speaks of this last act of the kind :

Postremo, Decembri jam mense; Lastly, when it was now the month of December-though the business of the procuration itself had very possibly begun as early as the Roman March or April.

It is therefore a singular coincidence that the eclipse should have happened on the xiii Kalendas Februarias, the 18th of the first month in this year, and that this ceremony of the Lectisternium and the Saturnalia, among the other means and modes of procuring it, should have been fixed to the xiv Kalendas Januarias, the 17th of the last month in the same year. No one could say what connection so superstitious a people as the Romans might not have imagined between these two things. The coincidence at least is something remarkable; and was very probably not the effect of chance.

The temple of Saturn indeed had been dedicated in B. C. 496 U. C. 258 Varro, and probably on this same day too, December 17 Roman, that yeary. But this testimony of Livy proves that the Saturnalia had not been permanently attached to the same day before this year B. C. 217 U. C. 537%. And this leads us to observe that, B. C. 217 being assumed as the actual date of this attachment, and the day on which the Saturnalia were actually celebrated this year to have been the xiv Kalendas Januarias, December 17 Roman, and thereby to have become their fixed and stated date ever after; there is a passage in Macrobius, which may possibly throw some light on these proceedings, and at this time. Novius enim probatissimus Atellanarum scriptor ait,

Olim exspectata veniunt septem Saturnalia.

Memmius quoque, qui post Novium et Pomponium diu jacentem artem Atellanarum suscitavit, Nostri, inquit, majores velut bene multa instituere, hoc optime; A frigore fecere summo dies septem Saturnalia a.

This implies that the proper seat of this festival in the natural year was later than the winter solstice; and in the regular calendar it might fall as early as December 16, and as late as January 15: and in the rectified years of the cycle

Nundinal Cycle, Vol. ii. 101. No iv.

z Cf. Vol. i. 453.

a Saturnalia, i. x. 240.

its proper date was December 28. In this very year, U. C. 537 B. C. 217, it fell on January 1 B. C. 216. And as the mean winter solstice at the same time was falling on December 25, and the true on December 24, (about noon for the meridian of Rome,) it follows that between the first Saturnalia, celebrated on the stated day in question, and the mean winter solstice, there was just an interval of seven days.

It could not be the meaning of Memmius, in the passage quoted from him by Macrobius, that the Saturnalia were appointed at first to last seven days; for that would have been contrary to the matter of fact, and to the history of the institution itself. It is much more probable that he intended to say that they were instituted at first seven days after the middle of winter: which would be literally true of this institution B. C. 217-216. Add to this that these first Saturnalia of December 17 Roman were preceded by a Lectisternium ad Ædem Saturni; and the rule of the Lectisternium from the first seems to have been to last seven days, and some seven days between two Nundinal days in sequencec. U. C. 537 B. C. 217 Dec. 8-Dec. 16 Roman was one complete Nundinal cycle. This Lectisternium might be celebrated December 9-15 both inclusive; beginning on the day after one Nundinal day and ending on the day before the next. And these seven days coming so near to the date of the first Saturnalia, Dec. 17, the same year, and being so closely connected with them, might give occasion to the tradition which certainly appears to have been handed down among the Romans, (as Macrobius' testimony provesd,) that even the first Saturnalia lasted seven days.

ii. Solar eclipse July 26 B. C. 216.

U. C. 538 Varr. U. C. 537 Cap. U. C. 535 Polyb. B. C. 216.

C. Terentius Varro

L. Æmilius Paullus ii.

b Vol. i. p. 453.

c Vol. ii. 113.

d Loco cit.

Maiæ
vii Juniæ July 3

May 4
June 2

mid. 216

vii

i Januariæ Jan. 14 mid. 216 v Apriles
ii Februariæ Feb. 12
iii Interkalares Mar. 7

iv Martiæ April 3

Kalendas Quinctiles July 26.

Pingré, eclipsed July 26 8.45 P. M. Paris B. C. 216.

This was the year of Cannæ. Among other prodigies, supposed to have preceded and prognosticated that event, Silius Italicus mentions an eclipse of the sun.

Quæsivit Calaber subducta luce repente
Immersus tenebris et terram et litora Sipus;
Obseditque frequens castrorum limina bubo:
Nec densæ trepidis apium se involvere nubes
Cessarunt aquilis: non unus crine corusco
Regnorum eversor rubuit letale cometes e.

Whatsoever may be thought of some of these prodigies, the first seems to be intended of something historical; of the fact of a sudden darkness, and greater than usual, observed at Sipus or Sipontum on the Sinus Urias, in the ancient Daunia, the site of which was near that of the modern Manfredonia in Italy f. For this is no doubt the locality intended; though called by its Greek name of Sipus, and not by its Latin one of Sipontum 5.

There were two solar eclipses, B. C. 216; one January 31 at 6.45 A. M. for the meridian of Paris according to Pingré. But that must have been much too early in the year to have been construed into a presage of the battle of Cannæ, fought in the month of September the same year, even if it had been visible for any part of Italy; which may be considered a doubtful point. The other, according to Pingré, was that of July 26, at 8.45 P. M. for the meridian of Paris and that would seem to have been too late to be visible at Sipontum.

We may collect however from the language of the above allusion that, if the darkness in question was the effect of

e Punica, viii. 631.

f D'Anville, i. 177.

g Cf. Polybius, x. 1 § 8. Pomponius Mela, ii. 4. p. 54. Pliny, H. N. iii. 16.

673. Livy, viii. 24: xxxiv. 45: xxxix. 23. Cicero, xvi. De Lege Agraria, ii. 27, 71: Ad Att. x. 7. Dio, xlviii. 27: 28. Servius, ad Æn. xi. 247.

a solar eclipse at the time, or an accompaniment of one, it must have happened towards the end of the day; and that the sun in fact must have gone down eclipsed, and sunset itself have been followed by an almost instantaneous darkness. For this appears to have been what the description intended by the land and the shore of the sea's being suddenly lost from view in a prodigious cloud of darkness. The external phenomena of this eclipse consequently must have been much the same as those of the eclipse of May 26, B. C. 715, at the death of Romulus, and as those of the eclipse of June 21 B. C. 400; both which happened under very similar circumstances, towards the end of the day: and sunset at that time too was followed by almost immediate darkness.

We have thought it worth while to calculate this ecliptic conjunction from our own Tables: and we find that for the meridian of Sipontum, (supposed to be 1h. 4m. 2 sec. east of that of Greenwich,) it happened July 26 at 21 h. 19 m. 28 sec. from midnight. And that may be assumed as about the middle of the eclipse. But the magnitude of this eclipse must have been something considerable; the mean distance of the sun from the ascending node being only about 5o 26′ or 27′, and the true only 3° 43′ or 44′′. It would consequently last three or four hours. We find from calculation also that apparent sunset on the same day, for the latitude of Sipontum, would take place about 19h. 23 m. meantime from midnight; 1h. 56 or 57 minutes before the middle of the eclipse. We may conclude then that the eclipse would be beginning at sunset; and that the sun would actually go down in a state of partial eclipse, and not improbably in the midst of a convulsion of the elements: which, combined with the effect of the eclipse, might make its disappearance be followed by almost instant darkness. It is manifest at least that, with an eclipse of such magnitude, beginning at sunset and attaining to its maximum about two hours afterwards, the ordinary duration of twilight on this evening must have been very much shortened.

The Roman date of this eclipse would be vii Kalendas Quinctiles: June 24 Roman. The battle of Cannæ was fought on the iv Nonas Sextiles h, next after, September 2 h A. Gellius, v. 17. Macrobius, Saturnalia, i. 16. 286.