The Vedanta Sutras of Badarayana, Commentary by Sankara (SBE38), tr. by George Thibaut  at sacred-texts.com
2. On account of (the Self) standing in a supplementary relation (to action), (the statements as to
the fruits of the knowledge of the Self) are arthavâdas, as in other cases, thus Gaimini opines.
As the Self, in consequence of its being the agent, stands in a supplementary relation to action, the knowledge of the Self also is connected with action through the mediation of its object, analogously to the case of the sprinkling of the rice-grains with water; hence as the purpose of the knowledge of the Self is understood thereby, the statements of the text about the fruits of that knowledge are mere arthavâdas. Such is the opinion of the teacher Gaimini 1. The case is analogous to that of other textual statements as to the fruits of certain materials, samskâras and works; which statements have likewise to be understood as arthavâdas. Cp. the passage, 'He whose sacrificial ladle is made of parna-wood hears no evil sound;' 'By anointing his eye he wards off the eye of the enemy;' 'By making the prayâga and anuyâga-oblations he makes an armour for the sacrifice, an armour for the sacrificer so that he overcomes his enemies 2.'--But how can it be supposed that
the knowledge of the Self which the text does not exhibit under any special heading can enter into sacrificial action as a subordinate member, without the presence of any of the means of proof--general subject-matter and so on--which determine such subordinate relation?--The pûrvapakshin may reply that the knowledge of the Self enters into sacrificial action through the mediation of the agent, on the ground of the means of proof called vâkya (sentence; syntactical unity) 1. But this we deny because in the present case 'sentence' has no force to teach the application (of the knowledge of the Self to the sacrifices, as a subordinate member of the latter). Things which the text states under no particular heading may indeed be connected with the sacrifice on the ground of 'sentence,' through some intermediate link which is not of too wide an application 2; but the agent is an intermediate link of too wide an application, since it is common to all action whether worldly or based on. the Veda. The agent cannot therefore be used as a mediating link to establish the connexion of the knowledge of the Self with the sacrifice.--Your objection is not valid, the pûrvapakshin replies, since the knowledge of a Self different from the body is of no use anywhere but in works based on the Veda. For such knowledge is of no use in worldly works, in all of which the activity may be shown to be guided by visible purposes; with reference to Vedic works, on the other hand, whose fruits manifest themselves only after the death of this body no activity would be possible
were it not for the knowledge of a Self separate from the body, and such knowledge therefore has its uses there.--But, another objection is raised, from attributes given to the Self, such as 'free from sin,' and the like, it appears that the doctrine of the Upanishads refers to that Self which stands outside the samsâra and cannot therefore be subordinate to activity.--This objection too is without force; for what the Upanishads teach as the object of cognition is just the transmigrating Self, which is clearly referred to in such terms as 'dear' (Bri. Up. II, 4, 5). Attributes such as being free from sin, on the other hand, may be viewed as aiming merely at the glorification of that Self.--But in more than one place Brahman, the cause of the world, which is additional to the transmigrating Self and itself not subject to transmigration has been established, and the Upanishads teach that this very Brahman constitutes the real nature of the transmigrating Self!--True, that has been established; but in order to confirm that doctrine, objections and their refutation are again set forth with reference to the question as to the fruit (of the knowledge of the Self).
286:1 The contention of the pûrvapakshin--Gaimini--is that the knowledge of the Self has no independent fruit of its own, because it stands in a subordinate relation to sacrificial action. This relation is mediated by the Self--the object of knowledge--which is the agent in all action, and therefore itself stands in a subordinate relation to action. By learning that his Self will outlive the body the agent becomes qualified for actions, the fruit of which will only appear after death. The qualification the Self thus acquires is analogous to that which the rice-grains acquire by being sprinkled with water; for only through this latter act of ceremonial modification (or purification, samskâra) they become fit to be used in the sacrifice.--As the knowledge of the Self thus has no independent position, it cannot have an independent fruit of its own, and consequently the passages which state such fruits cannot be taken as 'injunctions of fruits,' but merely as arthavâdas, making some additional statement about the fruit of the sacrificial actions to which the knowledge of the Self is auxiliary.
286:2 The material, i.e. the ladle made of parna-wood, is auxiliary to the sacrifice, and the fruit which the text ascribes to it (viz. hearing no evil sound) therefore has to be viewed as a fruit of p. 287 the entire sacrifice. Analogously in the case of the samskâra--the anointing--which fits the sacrificer for performing the sacrifice, and in the case of the prayâgas and anuyâgas which are merely subordinate members of the darsapûrnamâsa.
287:1 The entire Veda constituting an extended syntactical whole, in which the agent is the same.
287:2 Thus the quality of being made of parna-wood is connected with the sacrifices on the ground of the vâkya implied in 'yasya parnamayî guhûr bhavati,' because here we have as an intermediate link the guhû, i.e. a special implement which is used at sacrifices only, and therefore is not of too wide an application.