Kapila is answering his mother's question about Prakṛti and Puruṣa, and he gives the tattva count as a formula:
The learned know the brahman as the effects of Pradhana, with an aggregate of twenty-four attributes — five, five and fourteen.1
Then the itemisation:
There are five mahabhutas — earth, water, fire, air and space. It is my view that their tanmatras, smell and the others, are also similar in number. There are ten organs of sense and these are said to be ears, skin, eyes, tongue, nose, the organ of speech, hands, feet, organ of generation and the anus as the tenth. Those that are internal are manas, buddhi, ahamkara and chitta. The difference between these four is noticed in the way they function.1
Five elements. Five subtle elements. Ten organs. Four internal. Twenty-four.
And then the addition that makes the scheme unusual:
These are enumerated as the attributes of the brahman and I have spoken about their arrangement. Time is the twenty-fifth.2
Manas, buddhi, ahaṃkāra, citta — and the text says the difference between them is noticed in the way they function, which is a promise to say what the functions are. Debroy supplies them:
The mind creates doubt, intelligence leads to conclusions, ego leads to pride and consciousness results in remembrance.3
Four operations, one each, and they are not degrees of the same thing.
Manas produces doubt — the faculty that opens a question. Buddhi produces conclusions — the faculty that closes one. Ahaṃkāra produces pride — the faculty that attaches an outcome to a self. Citta produces remembrance — the faculty that keeps it.
Doubt, decision, ownership, memory. That is a complete cycle of a thought, and the four are distinguished by where in the cycle each operates rather than by how refined each is.
Note what the ordering implies. Doubt comes first, which makes uncertainty the normal opening state of the mental apparatus rather than a failure of it — nothing here treats manas as defective for producing questions. And ownership sits third, after the conclusion and before the memory, which means a thought is claimed by a self only once it has resolved and before it is stored.
So what gets remembered is already owned. There is no stage at which a settled thought exists unattached, and that is a claim about why memory is so hard to hold loosely: by the time citta has it, ahaṃkāra has already been through.
The Bhāgavata gives a tattva scheme at 2(9) as well, and the apparatus itemises it:
The four are Prakriti, Purusha, Mahat and Ahamkara. The sixteen are the five organs of perception (eyes, ears, nose, tongue, skin), the five organs of action (hands, legs, stomach, genital organs, organ of excretion), five elements and the mind. The five are the tanmatras. These twenty-five are his tattvas.4
Four plus sixteen plus five. Twenty-five.
Both schemes total twenty-five. They contain different things.
Lay the memberships side by side and the differences are structural, not cosmetic.
Prakṛti, Puruṣa and Mahat are tattvas in the first scheme and absent from the second's twenty-four. At 2(9) they are three of the four foundational items. At 3(26) the twenty-four are effects of Pradhāna — so Prakṛti is what the list is derived from, not an item on it.
The internal organ is one in the first and four in the second. At 2(9) the mind is a single item among the sixteen. At 3(26) manas, buddhi, ahaṃkāra and citta are four distinct faculties with four distinct functions.
Ahaṃkāra changes category entirely. It is one of the foundational four at 2(9) and one of the internal four at 3(26) — moved from the ground of the scheme into its contents.
And time appears only in the second. At 2(9) the twenty-five are complete without it. At 3(26) the twenty-four are complete and time is added as a twenty-fifth of a different kind.
So the two schemes agree on a number and on almost nothing else. The elements and the tanmātras are common to both; the organs are counted as ten in one and split into two fives in the other; everything at the top of the scheme has moved.
And the organs do not match either, once the two fives are unpacked.
At 3(26) the ten run: ears, skin, eyes, tongue, nose — then the organ of speech, hands, feet, organ of generation, anus.1
At 2(9) the five organs of action are given as hands, legs, stomach, genital organs, organ of excretion.4
Stomach where the other has the organ of speech. That is not a variant name for one thing; they are different organs with different functions, and speech is the standard first item in the classical list of karmendriyas.
So the disagreement reaches down into the itemisation, not merely the grouping. Two lists, both said to contain five organs of action, sharing four members.
Which makes the matching total more remarkable rather than less. A scheme can differ at the top — Prakṛti in or out, ahaṃkāra foundational or contained — and differ at the bottom, in which specific organ occupies a slot, and still arrive at twenty-five, because twenty-five was never derived. It was the target.
The formula Kapila uses says so in its own shape: five, five and fourteen. Not twenty-four, which he then states — a grouping, given first, that a reader has to add up. Fourteen is itself a composite of ten and four and is never broken out in the formula. The count is presented as an arithmetic that resolves, and what it resolves to was fixed before any of the items were named.
Register: this is a comparison of the itemisations as they stand in the translation, one in the text and one in an endnote. Neither passage acknowledges the other and no note connects them. What can be shown is the two lists and the fact that both arrive at twenty-five.
The coincidence is the interesting part.
Twenty-five is the classical Sāṅkhya total, and it is fixed. Any scheme presented as Sāṅkhya has to land on it, because the number is what identifies the system.
So the number is the constraint and the membership is the variable. Two different arrangements of the tattvas, from two different sources or two different moments, both audited against the same total and both passing.
That is the opposite of the volume's usual failure mode. Elsewhere the membership holds and the count drifts — seven sheaths and eight, four ignorances and five.5 Here the count is immovable and the contents change underneath it.
Which is a more consequential kind of disagreement, because a stated total that is always correct provides no signal that anything has moved.
The four-way split is the scheme's substantive contribution, and it is worth taking seriously rather than treating as elaboration.
A single manas has to account for everything mental: perceiving, deciding, owning, recalling. That is a lot for one item, and it makes the psychology thin — anything that goes wrong internally goes wrong in the mind.
Four faculties with four operations permit a diagnosis. A failure of buddhi is an inability to conclude. A failure of citta is an inability to retain. And ahaṃkāra's function being pride — attaching outcomes to a self — makes the ego a stage in ordinary cognition rather than a moral defect.
That last point matters for the volume's wider claims. The three kinds of ahaṃkāra at the same chapter's opening produce the senses, the elements and the mind;6 here ahaṃkāra produces pride as its normal operation. In neither place is it something a person could simply not have.
The twenty-fifth is added with a definition that puts it outside the series:
Time is the twenty-fifth. It is said that the power of Purusha that causes fear is time.2
Every other item is a component — an element, a faculty, an organ. Time is described as a power of Puruṣa, and specifically as the power that produces fear.
So the scheme's last item is not made of the same material as the first twenty-four, and the volume develops this immediately: he is in the form of Purusha inside and is in the form of time on the outside.7
The point is worked at length elsewhere.8 What matters for the count is that the twenty-fifth is a different kind of entry, which means twenty-five is not really a total. It is twenty-four plus one.
You have two versions of a framework — a competency model revised, a taxonomy from two teams, a spec and its successor — and both have the same number of items.
The matching total is the trap and it is worth naming.
Diff the membership, not the count. A count that agrees tells you nothing, and it actively suppresses the question. Both of these schemes say twenty-five, and one contains Prakṛti, Puruṣa and Mahat while the other treats those as what the list derives from.
Watch for items that changed level. Ahaṃkāra is foundational in one and contained in the other. That is the most consequential kind of move and the hardest to see, because the item is present in both and looks unchanged. Anything that migrates between the ground of a scheme and its contents has changed what the scheme is claiming.
And check whether one version splits a single item. Manas becomes four faculties with four operations. A split of that kind is usually the version's actual contribution, and it is invisible in a total that was fixed in advance.
The general rule: where a total is externally constrained — by convention, by a standard, by what the number is known to be — the total carries no information and the membership carries all of it. Any audit that checks the count has checked the one thing that could not have been wrong.
The strongest evidence that the twenty-five is a fixed constraint rather than a result is that both schemes reach it by incompatible routes. One counts Prakṛti, Puruṣa and Mahat among the tattvas; the other treats the twenty-four as effects of Pradhāna and adds time.
Tension — the volume gives two tattva schemes with the same total and substantially different memberships, one in the text at 3(26) and one in an endnote at 2(9). Neither passage acknowledges the other. Preserved unresolved.
Tension — ahaṃkāra is foundational at 2(9) and a contained faculty at 3(26). The change of level is never marked. Preserved.
Open question. The fourfold internal organ with its four functions is supplied partly by the text and partly by endnote 1101. Is the functional assignment — doubt, conclusion, pride, remembrance — attested in the Sanskrit, or is Debroy importing it from a commentarial source?
This vault's Sāṅkhya material treats the twenty-five tattvas as the system's defining commitment — the count being what distinguishes classical Sāṅkhya from its neighbours, and the membership being settled.
The membership is not settled here. The Bhāgavata supplies two arrangements, and the one Kapila delivers is the one with a fourfold internal organ and time as the twenty-fifth — neither of which is standard.
Which is consistent with what the volume does to Sāṅkhya generally. Its own avatāra roster claims Kapila as the fifth descent, declares that over time, this knowledge has been lost, and then supplies a replacement teaching in Kapila's own mouth.9 This scheme is that replacement, and it does not match the scheme the same volume gave four chapters earlier.
With Hierarchy of Influence Factors. That page works a ranked model of what actually moves behaviour — the value of the model lying in the ordering, since knowing the factors without their relative weight yields no decision.
The tattva schemes are the opposite kind of object: enumerations whose ordering is derivational rather than evaluative, and whose totals are fixed by convention.
What the pairing produces: an influence hierarchy fails visibly when its ordering is wrong, because someone applies it and the predicted lever does not move. That failure is the model's error-correction. A tattva scheme has no such mechanism — nothing in practice depends on whether ahaṃkāra is foundational or contained, so no application ever contradicts either version, and both can persist indefinitely in the same corpus. Which locates the condition under which enumerations drift: a classification is stable only where something downstream breaks when it is wrong. Where the list is descriptive and complete and nothing is computed from it, the count will be maintained (because it is quoted) and the membership will not (because it is not). That is exactly the pattern here, and it is the pattern in most internal taxonomies that nobody has ever had to act on.
Sharpest implication. Both schemes are correct about the number and neither can be checked by it. Where a total is fixed by convention, it becomes the one part of a framework guaranteed not to reveal a change — and the parts that did change, including an item moving from the ground of the system into its contents, pass unremarked in a text that quotes the total twice.
Generative questions.