The Vedanta Sutras of Badarayana, Commentary by Sankara (SBE38), tr. by George Thibaut  at sacred-texts.com
25. Because the matter (of certain mantras) such as piercing and so on is different (from the matter of the approximate vidyâs) (the former have not to be combined with the latter).
At the beginning of an Upanishad of the Âtharvanikas the following mantra is recorded, 'Pierce him (the enemy) whole, pierce his heart: crush his veins, crush his head; thrice crushed,' &c. At the beginning of the Upanishad of the Tândins we have the mantra, 'O God Savitar, produce the sacrifice.' At the beginning of that of the Sâtyâyanins, 'Thou hast a white horse and art green as grass,' &c.; at the beginning of that of the Kathas and the Taittirîyakas, 'May Mitra be propitious to us and Varuna,' &c. At the beginning of the Upanishad of the Vâgasaneyins we have a Brâhmana-passage about the pravargya-ceremony, 'The gods indeed sat down to a sattra;' and at the beginning of that of the Kaushîtakins there is a Brâhmana-passage about the agnishtoma, 'Brahman indeed is the Agnishtoma, Brahman is that day; through Brahman they pass into
[paragraph continues] Brahman, immortality those reach who observe that day.'--The point to be inquired into with reference to all these mantras and the sacrifices referred to in the Brâhmana-passages is whether they are to be combined with the vidyâs (contained in the Upanishads) or not.
The pûrvapakshin maintains that they are so to be combined, because the text exhibits them in proximity to the Upanishad-portions of the Brâhmanas whose chief contents are formed by the vidyâs.--But we do not observe those mantras and sacrifices to be actually enjoined as subordinate members of the vidyâs!--True, but in spite of this we, on the ground of proximity, infer them to be connected with the vidyâs. For we have no right to set aside the fact of proximity as irrelevant as long as an inference can be established on it.--But we are unable to see that the mantras have anything to do with the vidyâs, and how can it be assumed that ceremonies, such as the pravargya which scripture enjoins with reference to other occasions, sacrifices, and so on, stand in any relation to the vidyâs!--Never mind, the pûrvapakshin replies. In the case of mantras we can always imagine some meaning which connects them with the vidyâs; the first mantra quoted, e.g. may be viewed as glorifying the heart. For the heart and other parts of the body are often represented, in the vidyâs, as abodes of meditation, and hence mantras glorifying the heart, &c., may appropriately form subordinate members of those vidyâs. Some mantras, moreover, we clearly see to be enjoined with reference to vidyâs, so, e.g. the mantra, 'I turn to Bhûh with such and such' (Kh. Up. III, 15, 3). Sacrificial acts again may indeed be enjoined in connexion with other occasions; yet there is no reason why they should not also be applied to the vidyâs, just as the offering called Brihaspatisava is a subordinate part of the Vâgapeya-sacrifice 1.
To this we make the following reply. The mantras and
ceremonies mentioned cannot be drawn into connexion with the vidyâs, 'because their matter, such as piercing the heart, &c., is different (from the matter of the vidyâs),' and therefore cannot be connected with the latter.--But has it not been said above that the mantras may be connected with the meditations enjoined in the vidyâs, on the ground of their coming of use in meditations on the heart, &c.?--The mantras, we reply, might be so employed, if their entire contents were glorification of the heart, and the like; but this is by no means the case. The mantra first quoted, e.g. clearly expresses hostility to somebody, and is therefore to be connected, not with the vidyâs of the Upanishads, but with some ceremony meant to hurt an enemy. The mantra of the Tândins again, 'O God Savitar, produce the sacrifice,' indicates by its very words that it is connected with some sacrifice; with what particular sacrifice it is connected has to be established by other means of proof. Similarly other mantras also--which, either by 'indication' (liṅga), or 'syntactical connexion' (vâkya), or some other means of proof, are shown to be subordinate to certain sacrificial actions--cannot, because they occur in the Upanishads also, be connected with the vidyâs on the ground of mere proximity. For that 'proximity,' as a means of proof regarding the connexion of subordinate matters with principal matters, is weaker than direct enunciation (Sruti), and so on, is demonstrated in the former science (i.e. in the Pûrva Mîmâmsâ) under III, 3, 14. Of sacrificial works also, such as the pravargya, which are primarily enjoined with reference to other occasions, it cannot be demonstrated that they are supplementary to vidyâs with which they have nothing in common. The case of the Brihaspatisava, quoted by the pûrvapakshin, is of an altogether different kind, as there we have an injunction clearly showing that that oblation is a subordinate member of the Vâgapeya, viz. 'Having offered the Vâgapeya he offers the Brihaspatisava.' And, moreover, if the one pravargya-ceremony has once been enjoined for a definite purpose by a means of proof of superior strength, we must not, on the strength of an inferior means of proof, assume
it to be enjoined for some different purpose. A proceeding of that kind would be possible only if the difference of the means of proof were not apprehended; but in our case this latter possibility is excluded since the relative strength and weakness of the various means of proof is fully apprehended (on the ground of the conclusions arrived at in the Pûrva Mîmâmsâ).--For these reasons the mentioned mantras and acts are not, on the ground of mere textual collocation, to be viewed as supplementary to the vidyâs of the Upanishads. To account for the fact of their textual collocation with the latter we must keep in view that the mantras, &c. as well as the vidyâs have to be studied, &c. in the woods.
223:1 The Brihaspatisava, although enjoined with special reference to him who is desirous of Brahmavarkas, is yet at the same time a subordinate part of the Vâgapeya-sacrifice. Cp. Pû. Mî. Sû. IV, 3, 29.