^{page 213 note 1} I use the term water-clock with some hesitation, because it seems to give the idea of a dial with hands moved by wheels and cords or chains worked by water, whereas there was nothing of that kind in the Indian appliance: but it is difficult to find any other suitable term. The Greek term clepsydra is perhaps intrinsically nearer the mark, since it seems to mean literally “something from which water slips away in a thievish or elusive manner”, and to have denoted originally an appliance from which water trickled out: but it, too, in the later developments connotes a machine worked by water; also, when translated, it is rendered by “water-clock”; and so it is not really any better than the plain English term.

A general Indian term for the appliance was jala-yantra, ‘the water-instrument’, with any synonym of it: thus, Varāhamihira mentions it incidentally as ambu-yantra in his Bṛihat-Saṁhitā, 2. 3; and similarly the Sūrya-Siddhānta, chapter 13, speaks of it as tōya-yantra in verse 21, but in verse 23 calls it kapālaka, ‘the cup or bowl’. Other names of it were ghaṭl and ghaṭikā, ‘the water-jar’, and ghaṭī-yantra. It seems to be known now as kaṭōrā, kaṭōrī, ‘the cup, bowl, or dish’.

^{page 214 note 1} A notice, with details of size, etc., of a smaller copper bowl of this class from Ceylon has been given by MrSmith, Reginald in an interesting paper, published in 1907 in the Proceedings of the Society of Antiquaries of London, 2nd series, vol. 21, pp. 319–33Google Scholar, which is directed to showing that the British Druids had some appliance of the same kind from at least the first century b.c. Beyond that, I can only find a short reference to an Indian bowl by Gilchrist, , in a paper on “Hindustanee Horology”, in As. Res., vol. 5 (1798), p. 87.Google Scholar

^{page 214 note 2} As far as I can find, this has been recognized only by Dvivedi, S. (see p. 215 below)Google Scholar and Shamasastri, R. (see p. 219).Google Scholar I must correct a mistake which I made in JRAS, 1914, p. 174Google Scholar: in line 14 cancel “floating”, and in line 16 for “into” read “out of”.

^{page 215 note 1} See his Bhāratīya-Jyōtiḥśāstra or “History of Indian Astronomy,” Poona, 1896, p. 78.Google Scholar

^{page 215 note 2} See his Jyautisham, Benares, 1908, pp. 10, 39Google Scholar: this book embodies the substance of lectures delivered in previous years.

^{page 215 note 3} See his Jyōtisha-Vēdāṅga, Allahabad, 1907, p. 11.Google Scholar

^{page 215 note 4} The text is the current one, which may be found in any printed copy, with only the obviously proper corrections of pañchāśat-palam = āḍkakam for °palamāshakam, and āḍhakāt for māshakāt. About the third pāda see the next note.

^{page 215 note 5} Perhaps we might emend kumbhakō into kumbhakē, and so have:— “… from the āḍhaka there should be measured out a drōṇa in a water-jar.” But it seems desirable to take the standing text with as little alteration as is unavoidable.

Varāhamihira used the expression pañchāśat-palam = āḍhakam in his Bṛihat-Saṁhitā, 23. 2. The chapter deals with the prognostics for the rainfall, which were to be based on the quantity falling under the successive nakshatras coming just after the full-moon of Jyaishṭha; and he tells us that the fall was to be caught in a basin a cubit in diameter, and was to be measured out by “the āḍhaka of fifty palas”, and states in drōṇas the quantity which should normally fall under each nakshatra. In commenting on that, Bhaṭṭōtpala has quoted from some unspecified source:— pañchāśat-palam = āḍhakaṁ chaturbhir = āḍhakair = drōṇaḥ. Dikshit, conjectured (loc. cit., p. 215 above)Google Scholar that Bhaṭṭōtpala quoted the second and third pādas of this verse of the Jyōtisha-Vēdāṅga, and so has given what must be taken as the real original text of the third pāda. That is quite possible: but it involves a rather violent correction of the current text; and it does not follow of necessity, because Bhaṭṭōtpala may have quoted from some metrical work dealing with the measures of capacity (compare samples given farther on).

^{page 216 note 1} The text is the current one, which is correct just as it stands.

^{page 216 note 2} The text is the current one, which needs no correction: it is only to be noted that the Ṛig verse has viparyastau in the place of the viparyāsaḥ of the other.

^{page 217 note 1} The effect of this is that the length of the daytime ranged from 12 muhūrtas, = 9 hrs. 36 min., at the winter solstice to 18 muhūrtas, = 14 hrs. 24 min., at the summer solstice. Compare Divyāvadāna, , p. 642Google Scholar, lines 18–21; and Vishṇu-Purāṇa, , 2. 8. 31–7.Google Scholar The statement gives, according to the mean or uniform time to which it was adapted, very nearly 47 seconds as the time by which the sun rose earlier or later day by day, and 4.48 a.m. and 7.12 a.m. as the earliest and latest times of sunrise. As has been pointed out by previous inquirers, the result marks approximately the locality in which the rule was framed. Dikshit, (op. cit., p. 90)Google Scholar arrived at the close limits of lat. 34° 46′ to 55″. Barhaspatya, (op. cit., p. 32)Google Scholar has proposed more judiciously the rough wider limits of 32° 30′ to 42° 30′.

^{page 218 note 1} See the editors' remark on p. 655.

^{page 218 note 2} Or correct drōṇaḥ into drōṇaṁ, or ēkaṁinto ēkaḥ, and say “one drōṇa of water.”

^{page 218 note 3} There is perhaps something corrupt in the reading suvarṇa-mātram = upari. But we cannot doubt that the meaning is that the weight of the gold was to be a suvarṇa.

^{page 218 note 4} It may be noted, in passing, that the term rātri-divasa, ‘night-and-day’, reminding us of the Greek nycthemeron, is a rather peculiar one for India, where the day has always run from sunrise, not from sunset. Except in the indeclinables naktaṁdivasam and rātriṁdivam, ‘by night and day’, which are taught by Pāṇini, 5. 4. 77, and may be due to euphonic considerations, nothing matching it seems to be found in the Brāhmaṇical books: their terms are ahō-rātra, dina-rātri, dyu-niśa, etc., ‘day-and-night’, with very rarely ahanī, lit. ‘the two days’, which seems to be chiefly Vēdic, but is found in the Bhāgavata-Purāṇa, 3. 11. 10, 12. The Divyāvadāna has rātri-divasa again on p. 642Google Scholar, line 23, and rātriṁ-divasa on p. 642Google Scholar, line 17, and p. 643, line 4: but it presents the customary ahō-rātra on p. 644Google Scholar, lines 5, 12, 13, and p. 645, line 5.

^{page 219 note 1} The Kauṭilīya uses two forms, nāḍikā and nāḷikā, with a preference for the latter.

^{page 219 note 2} See p. 133 of his translation of the whole work, which will be published shortly brought together in one volume.

^{page 220 note 1} My references are:—(1) The Vāyu, 100. 219b to 221a; text edited and printed in 1905 in the Ānandāśrama Sanskrit Series, Poona.

(2) The Vishṇu, 6. 3. 7 to 9a; text, with the commentary of Ratnagarbha, printed in 1860 at the Vṛittadīpa Press, Bombay: and text, with the commentaries of Vishṇuchitta and Śrīdhara, printed in 1910 at the Śrī-Veṅkaṭēśvar Press, Bombay.

(3) The Brahma, 231. 7 to 9a; text edited and printed in 1895 in the Ānandāśrama Sanskrit Series, Poona.

(4) The Bhāgavata, 3. 11. 9; text printed in 1905 at the Nirṇayasāgar Press, Bombay. This statement is particularly imperfect, and also mixed: the first three pādas of the verse belong to the ancient form of the water-clock, but the fourth pāda to the later form. It is the more surprising because this Purāṇa is very archaic in some of its astronomical passages.

I am indebted to Mr. Pargiter for the references to the Vāyu and Brahma, and to Dr. Barnett for drawing my attention to Vishṇuchitta's very useful commentary on the Vishṇu: Śrīdhara and Ratnagarbha are quite misleading in this matter.

^{page 220 note 2} We infer from the Vishṇu and Brahma that in the Vāyu, also, at some time or another, the two verses quoted by Vishṇuchitta must have had before them the line Nāḍikā tu pramāṇēna, etc., and after them the line Nāḍikābhyām = atha dvābhyāṁ, etc. These two lines, which I give in brackets, are not in the printed text of the Vāyu, and did not come within the scope of Vishṇuchitta's quotation. I take it that, along with them, the two verses give what was once the standard text of the Purāṇas in this matter.

^{page 221 note 1} The commentators carefully explain ardha-trayōdaśa by sārdha-dvādaśa; and Vishṇuchitta and Srīdhara give grammatical explanations of the term.

^{page 222 note 1} The Divyāvadāna has māsaka as its form of māshaka.

^{page 222 note 2} The text and translation have vārī: but the real word is certainly khārī.

^{page 222 note 3} Colebrooke's translation, edited, with text and notes, by Banerji, H. Ch., Calcutta, 1893.Google Scholar

^{page 223 note 1} We may also compare, from Western India, Mahāvīra's Gaṇitasārasaṁgraha (of the period a.d. 815–77), text and translation, with notes, by Rangacharya, M., Madras, 1912Google Scholar, chapter 1, verses 36–8. This tells us that 4 shōḍaśikās = 1 kuḍaha (v.l. kuḍava), 4 kuḍahas (v.l. as before) = 1 prastha, 4 prasthas = l āḍhaka, 4 āḍhakas = l drāṇa, 4 drõṇas = l mānī, and 4 mānīs = l khārī. Except for introducing the mānī, this statement matches those of the Kauṭilīya and the Līlāvatī from the kuḍumba, kuḍava, to the khārī. The reading kuḍava seems better than kuḍaha.

^{page 223 note 2} The term for the piercing-tool in the Divyāvadāna, (p. 218 above)Google Scholar is śalākā, ‘a rod, pin, needle’. Elsewhere, in some of the later passages, use is made of sūchī, sūchī, ‘a needle’, and nala: this last word seems to be used mostly in the sense of ‘a hollow reed or stalk, a tube’; but there was obviously nothing hollow about the instrument that was used for piercing the water-jar.

According to the same work, the instrument might be either round or square: at least, I do not see what other meaning than ‘square’ may be given to chatur-asrā. One would think, however, that it would ordinarily be round.

^{page 225 note 1} See JRAS, 1912, p. 233.Google Scholar

^{page 225 note 2} See, e.g., Whitaker's Almanack, 1915, p. 450.Google Scholar

^{page 227 note 1} As checked by another process, namely, l drōṇa = 364½ cubic inches = ·2109375 cubic foot × 62·321 lb. (weight of one cubic foot of water: see Whitaker, p. 451)Google Scholar ÷ 200 palas, the amount by which l pala exceeds 1 oz. is ·001666875 oz.

^{page 227 note 2} These seeds, which come also from the West Indies, are probably familiar to my readers. Chambers, 's Encyclopædia, vol. 8, p. 382Google Scholar, says about them:— “Prayer Beads, a name given to the polished seeds of a West Indian leguminous plant, Abrus precatorius or Wild Liquorice, formerly much used for stringing into rosaries, necklaces, etc.”

^{page 227 note 3} JASB, vol. 33, p. 260Google Scholar, note 28. Compare the tables in his Ancient Indian Weights (1874), p. 13Google Scholar: the weights are given there in “grs. Troy”: but that is not necessary, as the grain is the same whether for troy or for avoirdupois; it is in the ounce that the difference in the two scales comes in.

^{page 228 note 1} Num. Chron., new series, vol. 13, p. 197Google Scholar: I quote from p. 11 of a reprint bound in at the end of his Coins of Alexander's Successors.

^{page 228 note 2} Coins of Ancient India, p. 44, and see the tables on pp. 46, 47, 59.Google Scholar

^{page 228 note 3} Compare the Mānava-Dharmaśāstra, 8. 131 ff., and the Yājñavalkya-Smṛiti, , 1. 361 ff.Google Scholar, as quoted by DrBarnett, in his Antiquities of India, p. 206Google Scholar: for the Kauṭilīya, tables see p. 210.Google Scholar

^{page 228 note 4} Whitaker's Almanack, 1915, p. 462.Google Scholar

^{page 229 note 1} This period of 1½ hours seems at first sight a rather curious one for India, because ib is not measurable by any even number of nāḍikās or mnhūrtas, and the hour only became known in India about a.d. 350–400, and even then was taken over only as an astrological item: it is only in quite modern times that the hour has been adopted in India as a division of time for practical purposes. But the matter is cleared up when we reflect that each such period is half of the prahara or yāma, the “watch” of three hours, the one-fourth division of the daytime and the night, which must have been fixed in very ancient times, because, even before the invention of the gnomon or any other appliances, it could be determined during the daytime, quite closely enough, by simple eyeobservation of the position of the sun, on the horizon, half-way up, and at the zenith.

For the prahara as the eighth part of an entire day-and-night, see, e.g., Bhaṭṭōtpala under Bṛihat-Saṁhitā, 24. 10: and for the yāma as the fourth part of a day and of a night, see, e.g., the Bhāgavata-Purāṇa, 3. 11. 10. For the identity of the prahara and the yāma, see also Amarakōśa, (Bombay, 1896), p. 24Google Scholar, verse 6; and Abhidhānachintāmaṇi, verse 145.

^{page 230 note 1} We may also imagine some such arrangement as follows. Take four water-jars, A, B, C, and D, each fitted with some appliance, which can be removed at once when wanted, for plugging or covering the hole in the bottom. Put D aside, and place A, B, C one above the other on a stand; A, at the top, having in it the water for measuring a half or a full watch. At the given moment set A discharging water into B; and when B is full, set the latter discharging into C. Then remove A, and replace it by D, filled meanwhile with the proper quantity of water. When C is full, set D discharging into B; remove C and replace it by A, empty; and add to C any small quantity of water wanted to make up any loss caused by waste or evaporation, so that it will be ready to take its turn at the top. Compare a form of the Chinese water-clock mentioned in this Journal, new series, vol. 18 (1886), p. 14; namely, bronze vessels arranged in successive steps as on a staircase (and connected, it is to be presumed, by pipes and taps).