Kṛṣṇa is telling Arjuna what to do with Aśvatthāmā, who has murdered sleeping children, and the word he uses for him is not brahmana:
O Partha! You should not save this brahma-bandhu. Kill him.1
Debroy glosses the construction, twice in the volume, in almost the same words:
bandhu, meaning those who are brahmanas only in name.2
bandhu, one who is a brahmana only in name.3
Bandhu on its own means kinsman, relative, connection. Attached to a varṇa it does something else: it marks someone as related to the category without being in it. A brahma-bandhu is a brahmin's relative. Not a brahmin.
That distinction is the whole interest of the word, and it is easy to lose.
There are plenty of insults available. Brahma-bandhu is not one of them — it does not say Aśvatthāmā is wicked, and the sentence next to it already says that at length. What it does is reclassify him, quietly, by a single morpheme, while continuing to acknowledge the connection.
He remains attached to the category. He is a brahmin's son, a brahmin's kinsman, everything a brahmin is except a brahmin. And the word says all of that at once.
Which makes it a technology rather than an epithet. An insult attacks a person's standing. This adjusts their membership, and does so in a form that concedes the relation it is denying.
The construction is not confined to brahmins. Śṛṅgi, the boy about to curse Parīkṣit, uses the parallel form:
This is like a servant acting against his master, or a dog acting against the doorkeeper. Brahmanas have appointed kshatra-bandhus as the guards of houses. In that house, how does a doorkeeper deserve to eat from the same vessel?4
Kṣatra-bandhu — kṣatriya only in name. And the sentence around it is doing exactly the work the word does: reducing a king to a guard, a guard to a dog, and denying him a place at the table.
Later, in the Fourth Skandha, a queen mourning her husband uses the same word for the disorder his death permits:
This earth, with the ocean as a girdle, is terrified of bandits and kshatra-bandhus.5
Bandits and kṣatra-bandhus, paired. A ruler who is only nominally a ruler is placed alongside outright criminals, and no argument is offered for the pairing — the word has done it.
The most interesting instance reverses direction. Dakṣa, having been punished and restored, addresses the deity:
O illustrious one! You and Hara do not ignore brahma-bandhus, not to speak of those who are firm in their vows. From your own mouth you first created brahmanas who know about the brahman, for the sake of upholding learning, austerities and vows.6
He is not condemning anyone. He is making a claim about divine attention: even the nominal brahmins are not ignored.
So the word appears here inside a compliment, as the low end of a scale that runs upward to those firm in their vows. And the sentence immediately after grounds the whole scale in the mouth-origin of brahmins.
Which shows what the term is for. It is not a weapon that only points outward. It is a grade — a position on the varṇa scale, below the full member, available for use in condemnation, in mercy, or in self-description depending on who is speaking.
Register: four instances across the volume, two words, one glossed twice by the apparatus. No passage defines the term and no note explains its formation beyond only in name. What can be shown is its distribution and the range of speakers who reach for it.
Three properties, and they compound.
It is not falsifiable. Brahma-bandhu makes no factual claim that could be checked. Aśvatthāmā's parentage is not in dispute; what is being asserted is that the parentage does not amount to membership, and there is no test for that.
It cannot be answered in kind. You cannot rebut a suffix. Denying it requires asserting your own full membership, which is exactly the defensive move that makes the question look live.
And it concedes enough to be credible. A speaker who called Aśvatthāmā a śūdra would be lying and would be heard as lying. Brahma-bandhu grants every fact — the birth, the lineage, the education — and withholds only the standing, which is the one thing not made of facts.
Set the term against two things this volume says elsewhere and the awkwardness is immediate.
It says nine named peoples are purified by refuge with devotees.7 It says women, śūdras, Hūṇas and Śabaras become good in conduct once instructed and devoted.8 It has the whole Purāṇa narrated by a sūta who names his own birth as a disqualification.
Those passages loosen the boundary at its outer edge. Bandhu tightens it at its inner edge — policing not who may come in from outside, but who among the already-included is genuinely in.
And the second is where boundaries usually get enforced. A tradition that admits outsiders generously and grades insiders finely has not become open; it has moved where the sorting happens.
Debroy glosses the term twice and both glosses are the same four words: only in name.
He does not note that the construction recurs, does not connect the brahmin and kṣatriya forms, and does not remark that the same word appears in Dakṣa's mouth as a compliment. Each instance is handled where it sits.
That is the volume's established annotation behaviour and it has the same consequence here as elsewhere: a reader consulting the notes gets the meaning of the word four times and never learns that it is a system. The pattern is only visible to someone reading the whole and comparing — which is not how the text was used and not how the apparatus was built.
Someone in your field, company or scene gets described with a qualifier that concedes the category and withholds it — technically a designer, a founder in the loose sense, nominally senior, one of the so-called originals.
That is the bandhu construction, and it is worth recognising because it does not behave like criticism.
It is unanswerable by performance. Doing the job better does not address it, because the claim was never that the work is bad. Aśvatthāmā's problem is not that his brahmin credentials are weak.
It is contagious in one direction only. Once the grade exists, it can be applied to anyone in the category, and its existence makes full membership something that has to be maintained rather than something possessed. That is the actual effect — not the demotion of one person but the introduction of a scale.
And the tell is the concession. Real disputes about membership deny the facts. This construction grants them all and withholds standing anyway, which is what makes it credible and what makes it impossible to rebut.
The response that works, where any does, is to ask what the grade's criteria are — since only in name has none, and demanding them either produces criteria you can meet or exposes that there were none. Neither party in this volume ever asks.
The strongest evidence that bandhu is a grade rather than an insult is Dakṣa's use of it. The same word appears in condemnation, in the mouth of a boy stripping a king of standing, in a queen's lament pairing it with bandits, and in a devotee's praise of divine generosity. An insult does not survive that range; a position on a scale does.
Tension — the volume admits outsiders generously and grades insiders finely. The boundary-work moves from the outer edge to the inner one, and the volume never places the two side by side. Preserved unresolved.
Open question. Four instances, two compounds, one gloss repeated. Is bandhu a productive suffix in this text — could it attach to vaiśya or śūdra — or does it only appear where a claim to the top two varṇas is being adjusted?
This vault's material on varṇa treats the categories as birth-assigned and the disputes as being about mobility between them — whether conduct can move a person, whether birth is decisive.
Bandhu is orthogonal to that argument and, in a sense, dissolves it. It does not move anyone between categories. It introduces a second dimension: within a category, the difference between being it and being related to it. Aśvatthāmā has not been reclassified downward; he has been graded within his own varṇa.
The Reciter They Call Low-Born documents the volume's narrator naming his own birth as a disqualification. Read together, the two show the boundary being worked from both sides in the same book — a low-born man reciting scripture at the outer edge, and brahmins being graded against each other at the inner one.
With Demand for Purity. That page works the mechanism by which a group sorts its own members against an absolute standard — the standard being unattainable by design, so that everyone inside is permanently gradeable and nobody is ever securely in.
Bandhu is that mechanism compressed into a suffix. It creates a position below full membership, inside the category, with no stated criteria for escaping it.
What the pairing produces: the purity analysis is built around an explicit standard that members are measured against — the doctrine, the line, the code — and its critique proceeds by showing the standard is unmeetable. This case has no standard at all. Nothing anywhere says what makes someone a brahmin rather than a brahma-bandhu; the grade is asserted by morphology and never defined. Which locates something the purity frame does not need to see, because its cases always supply a yardstick: the sorting does not require one. A grammatical position below the category is sufficient, and it is in some ways stronger than an explicit standard, because an explicit standard can be met, argued about, or shown to be unmeetable. A suffix can only be applied or not applied, and the decision belongs entirely to whoever is speaking. The absence of criteria is not a weakness in the mechanism. It is the mechanism.
Sharpest implication. The word concedes every fact about a person's membership and withholds the membership itself, which makes it unanswerable by evidence or by performance — and its existence converts full standing from something possessed into something continuously assertable by other people. A single morpheme does what an entire apparatus of criteria would be needed to do, and it does it without ever stating a criterion.
Generative questions.