One page earlier the tradition established that initiation is singular and structural. One initiator. A seed you cannot read, unpacking over years. Not a course you can switch.
Then, in the same limb, it specifies the conditions under which you leave.1
And the sentence it opens with is doing careful work: there is no directive permitting one to renounce a bona fide guru. Notice what that leaves. Nothing has been said about a guru who is not bona fide — which is where the entire passage lives.
A guru is to be given up who:
Look at what is not on that list. No mention of error at the time of initiation. No mention of the disciple's disappointment, or of an unmet expectation, or of the relationship having stopped working.
Every item is about present condition and orientation — what the man has become, not what he got wrong. The word the passage uses twice is has become.
Which means the test is not retrospective. You are not being invited to re-examine whether you chose well. You are being asked what he is now.
Two verses, and they pull in different directions in a way worth noticing.
From the Mahābhārata: for a guru who is arrogant, who does not know what should and should not be done, and who has taken to a wrong path — abandonment is prescribed. Parityāgo vidhīyate.2
That's not permission. Vidhīyate is the language of injunction. Abandonment is enjoined.
And from the Hari-bhakti-vilāsa: one who accepts a mantra from a non-Vaiṣṇava guru goes to hell — and therefore should receive the initiating mantra again, from a Vaiṣṇava guru, according to the rules.3
So the second verse makes staying actively dangerous, not merely unproductive. Inaction is not neutral. You are not holding a lapsed connection; you are holding a mantra from a source that consigns you.
Here's the part I find genuinely difficult, and the book does not address it.
Punaś ca vidhinā samyag grāhayed-vaiṣṇavād guroḥ — again, properly, by the rules, receive it from a Vaiṣṇava guru.
Re-initiation. A second mantra, from a second initiator, by the same procedure.
Which sits directly against the account one page earlier: there is only one initiating teacher, because the mantra carries a single interior orientation in seed form and you cannot be running two.
Both claims are in the same limb of the same Text. And the reconciliation the tradition presumably intends — that the first was never a real initiation, so there is still only one — depends on a fact about the first teacher's condition at the time, which is precisely what the six conditions do not test. They test what he is now.
So the operative situation is: a man who was a genuine teacher and has since become one of the six things. On the tradition's own account you must leave and re-initiate. And there is now a second seed. The singularity claim and the exit clause cannot both be fully true, and the book runs them a page apart.
There's a separate warning earlier in the same discussion, and it is aimed at a different failure.
Some people, without thoughtful consideration, take shelter of an impersonalist renunciant, or of a prākṛta-sahajiyā — glossed in the book's own footnote as one who abandons the fundamental procedures of practice and imitates the symptoms of advanced devotees by a material display — or of an ignorant family teacher, a kula-guru. Having done so, they think they have obtained a genuine teacher, and they conclude that no practice and no company are needed.
The book's verdict: this kind of consideration is blind faith. Such a spiritual master simply hinders one's attainment of prema.4
Note the difference from the six conditions. This isn't decline; it's a mis-selection at the start, and its diagnostic sign is not the teacher's behaviour but the disciple's conclusion — that nothing further is required. The tell is in the student.
The Mahābhārata verse lists three grounds, and it is worth noticing which one it puts first.
Guror apy avaliptasya — even of a guru who is avalipta.
The book renders it as proud, which is right and slightly weak. Avalipta is from a root meaning smeared or anointed — coated in something. Its extended sense is self-satisfied, puffed, glossy with one's own standing. Not pride as a flash of arrogance, but pride as a film over the surface.
Then the other two: not knowing what should and should not be done, and having taken to a wrong path.
Read as a sequence, that's a progression rather than a list. The coating comes first. Discrimination fails next — which is what you would predict of someone whose self-regard has become the medium he sees through. And the wrong path is the outcome.
Which means the tradition's first-named ground for enjoined abandonment is not doctrinal error, not misconduct, and not cruelty. It is a condition of the man's relationship to himself — and it is the one the six conditions in the commentary echo when they name engrossed in sense enjoyment and lacking discrimination as separate items.
There's also something uncomfortable in the api — even of a guru. The verse is framed as an extension of a general principle to a case where you would expect immunity. Abandonment is enjoined for such a person even when that person is your teacher, which implies the ordinary rule already covered everyone else.
Something has changed and you have known it for about a year.
He speaks dismissively about other devotees now, in a way he didn't. There's a comfort around him that wasn't there. And when you raise something from the scriptures that cuts against a position he's taken, the response is not argument, it's irritation.
Run the list. Inimical to Vaiṣṇavas — arguably. Engrossed in enjoyment — you'd have to be honest about the comfort. Lacking discrimination — maybe. Inimical to scripture — that irritation is data.
What the list does not let you say is I've outgrown him, or I'm not getting what I need, or the relationship has run its course. None of those are on it. The exit conditions are about what he has become, and if none of them apply, the opening sentence applies instead: there is no directive permitting you to leave.
Now the harder half. The verse says abandonment is enjoined, not permitted, and that staying with a non-Vaiṣṇava mantra sends you down. So if the conditions do apply, drift is not an option available to you. The tradition has removed the comfortable middle — the quiet distance, the reduced contact, the not-quite-leaving that most people actually do.
And the third thing, which is the one nobody wants. Who adjudicates? The passage gives conditions and no procedure. You are the one applying a test whose outcome determines whether you may stop obeying the person you would ordinarily ask.
The structural problem here is worth stating plainly, because it recurs wherever exit clauses exist inside hierarchies.
The six conditions are reasonably objective. Hostility to devotees, absorption in enjoyment, and departure from the path of pure devotion are things other people can observe.
But the tradition supplies no adjudicator. There is no council, no appeal, no third party named. The disciple assesses.
And the disciple is, by the architecture of this same book, the party least equipped to assess. His discrimination is precisely what the relationship was supposed to develop. His judgement about what counts as departure from pure devotion depends on knowledge he was receiving from the person under assessment. And a substantial part of the surrounding literature is devoted to warning him that his own dissatisfaction is not evidence of anything.
Which produces a predictable distribution of outcomes. The exit clause will be under-used by the scrupulous, who distrust their own judgement exactly as they were taught to, and over-used by the aggrieved, who have a list to hang a grievance on. A specification without an adjudicator does not fail randomly; it fails along the lines of temperament.
I don't think the tradition is unaware of this. It's why the passage opens by closing the easy door — no directive permits leaving a genuine teacher — and why the whole weight falls on establishing that he isn't one.
Strongest support. Two scriptural warrants, both cited, and the stronger of the two uses injunctive language — abandonment is enjoined rather than allowed. And the six conditions are behavioural and present-tense rather than appeals to the disciple's satisfaction.
The tension the book does not close. Re-initiation is prescribed, and one page earlier the initiating teacher is singular because the mantra carries one interior orientation as an unreadable seed. The available reconciliation — that the first was never genuine — requires a judgement about the teacher's condition at the time of initiation, and the six conditions test his condition now. The case of genuine-then-fallen is the modal case and it is the one the reconciliation cannot cover.
A second tension. Accepting a mantra from a non-Vaiṣṇava is said to send one to hell. The disciple who did so in good faith, having applied the specification as carefully as he could, is on this account already consigned — and the remedy offered is procedural (be initiated again) rather than exculpatory. Nothing addresses the sincerity of the original error.
Open question. Who decides, and against what standard? The conditions are given and the adjudicator is not. In practice this is settled by whichever institution the disciple is inside, which means the exit clause's operation depends entirely on a body the passage never mentions.
The vault's material on teacher-transmission is largely written from the inside of a functioning relationship — what a teacher supplies that a text cannot, why the transmission requires a person.
This passage is the same tradition writing its failure procedure, and the change in register is total. No mention of what is transmitted, no mention of mood or seed or correspondence of nature. Six behavioural conditions, two verses, and a re-initiation instruction.
Which is worth noticing as a general feature of institutional literature: the exit clause is always written in a different vocabulary from the entry. Entry is described in terms of what passes between people. Exit is described in terms of observable conduct — because conduct is the only thing a rule can be written about, and a rule is what you need when the relationship it governs has stopped working.
An exit clause written by the same authority it permits you to exit, with the conditions specified and the adjudicator left out. Set this against Authority Construction and the Architecture of Belief. That page's mechanism is deference engineered structurally rather than claimed — authority located in an office, a text, a lineage, so that questioning it requires standing outside a frame you are inside.
The six conditions are that structure's release valve, and its design is revealing. The conditions are real and reasonably objective. The warrant is scriptural rather than personal, which is genuinely load-bearing — it means the teacher's own authority cannot override it. And no adjudicating body is named, which leaves the judgement with the person whose judgement the entire relationship was constituted to form.
Here's what the pair produces that neither gives alone. A constructed authority that provides no exit clause is brittle; one that provides an exit clause with no adjudicator is stable and self-selecting. Brittle systems break publicly, because the only available exit is a rupture, and ruptures create precedent and factions. Systems with an unadjudicated exit clause absorb the strain instead: the scrupulous stay, because they have been trained to distrust exactly the faculty the clause requires them to use, and the aggrieved leave under their own authority, taking their grievance with them and generating no institutional precedent because no institution ruled on anything.
That's a more durable arrangement than either total closure or a formal tribunal — and it isn't necessarily designed. It's what you get when you write honest conditions and don't specify who applies them.
Sharpest implication. The exit clause and the singularity claim cannot both be fully true, and the case that breaks them is the common one: a genuine teacher who has since become one of the six things. Re-initiation gives you a second seed, which the singularity argument says is impossible — and the tradition runs both claims a page apart without reconciling them.
Generative questions.
prākṛta-sahajiyā error, defined by imitation of advanced symptoms