The ancient Upanishads, i.e. those which occupy a place in the Samhitâs, Brâhmanas, and Âranyakas, must be, if we follow the chronology which at present is commonly, though, it may be, provisionally only, received by Sanskrit scholars, older than 600 B.C., i.e. anterior to the rise of Buddhism. As to other Upanishads, and their number is very large, which either stand by themselves, or which are ascribed to the Atharva-veda, it is extremely difficult to fix their age. Some of them are, no doubt, quite modern, for mention is made even of an Allah-upanishad; but others may claim a far higher antiquity than is generally assigned to them on internal evidence. I shall only mention that the name of Atharvasiras, an Upanishad generally assigned to a very modern date, is quoted in the Sûtras of Gautama and Baudhâyana^[32]; that the Svetâsvatara-upanishad, or the Svetâsvataranâm Mantropanishad, though bearing many notes of later periods of thought, is quoted by Sankara in his commentary on the Vedânta-sûtras^[33]; while the Nrisimhottaratâpanîya-upanishad forms part of the twelve Upanishads explained by Vidyâranya in his Sarvopanishad-arthânubhûti-prakâsa. The Upanishads comprehended in that work are:

The number of Upanishads translated by Dârâ Shukoh amounts to 50; their number, as given in the Mahâvâkyamuktâvalî and in the Muktikâ-upanishad, is 108^[35]. Professor Weber thinks that their number, so far as we know at present, may be reckoned at 235^[36]. In order, however, to arrive at so high a number, every title of an Upanishad would have to be counted separately, while in several cases it is clearly the same Upanishad which is quoted under different names. In an alphabetical list which I published in 1855 (Zeitschrift der Deutschen Morgenländischen Gesellschaft XIX, 137-158), the number of real Upanishads reached 149. To that number Dr. Burnell^[37] in his Catalogue (p. 59) added 5, Professor Haug (Brahma und die Brahmanen) 16, making a sum total of 170. New names, however, are constantly being added in the catalogues of MSS. published by Bühler, Kielhorn, Burnell, Rajendralal Mitra, and others, and I shall reserve therefore a more complete list of Upanishads for a later volume.

Though it is easy to see that these Upanishads belong to very different periods of Indian thought, any attempt to fix their relative age seems to me for the present almost hopeless. No one can doubt that the Upanishads which have had a place assigned to them in the Samhitâs, Brâhmanas, and Âranyakas are the oldest. Next to these we can draw a line to include the Upanishads clearly referred to in the Vedânta-sûtras, or explained and quoted by Sankara, by Sâyana, and other more modern commentators. We can distinguish Upanishads in prose from Upanishads in mixed prose and verse, and again Upanishads in archaic verse from Upanishads in regular and continuous Anushtubh Slokas. We can also class them according to their subjects, and, at last, according to the sects to which they belong. But beyond this it is hardly safe to venture at present. Attempts have been made by Professor Weber and M. Regnaud to fix in each class the relative age of certain Upanishads, and I do not deny to their arguments, even where they conflict with each other, considerable weight in forming a preliminary judgment. But I know of hardy any argument which is really convincing, or which could not be met by counter arguments equally strong. Simplicity may be a sign of antiquity, but it is not so always, for what seems simple, may be the result of abbreviation. One Upanishad may give the correct, another an evidently corrupt reading, yet it does not follow that the correct reading may not be the result of an emendation. It is quite clear that a large mass of traditional Upanishads must have existed before they assumed their present form. Where two or three or four Upanishads contain the same story, told almost in the same words, they are not always copied from one another, but they have been settled independently, in different localities, by different teachers, it may be, for different purposes. Lastly, the influence of Sâkhâs or schools may have told more or less on certain Upanishads. Thus the Maitrâyanîya-upanishad, as we now possess it, shows a number of irregular forms which even the commentator can account for only as peculiarities of the Maitrâyanîya-sâkha^[38]. That Upanishad, as it has come down to us, is full of what we should call clear indications of a modern and corrupt age. It contains in VI, 37, a sloka from the Mânava-dharma-sâstra, which startled even the commentator, but is explained away by him as possibly found in another Sâkhâ, and borrowed from there by Manu. It contains corruptions of easy words which one would have thought must have been familiar to every student. Thus instead of the passage as found in the Khândogya-upanishad VIII, 7, 1, ya âtmâpahatapâpmâ vigaro vimrityur visoko 'vigighatso 'pipâsah, &c., the text of the Maitrâyanîya-upanishad (VII, 7) reads, âtmâpahatapâpmâ vigaro vimrityur visoko 'vikikitso 'vipâsah. But here again the commentator explains that another Sâkhâ reads 'vigighatsa, and that avipâsa is to be explained by means of a change of letters as apipâsa. Corruptions, therefore, or modern elements which are found in one Upanishad, as handed down in one Sâkhâ, do not prove that the same existed in other Sâkhâs, or that they were found in the original text.

All these questions have to be taken into account before we can venture to give a final judgment on the relative age of Upanishads which belong to one and the same class. I know of no problem which offers so many similarities with the one before us as that of the relative age of the four Gospels. All the difficulties which occur in the Upanishads occur here, and no critical student who knows the difficulties that have to be encountered in determining the relative age of the four Gospels, will feel inclined, in the present state of Vedic scholarship, to speak with confidence on the relative age of the ancient Upanishads.