You start doing the thing every morning and it works. Six months in you are noticeably better at it, and you are also — and this arrives more slowly — noticeably more insufferable about it.
Not because you got worse as a person. Because the same discipline that built the skill built the identity that comes with the skill, and the second was watered by exactly the same water. You never decided to become someone who mentions it. It grew in the same bed, on the same schedule, and you were the one holding the hose.
Madhya 19 says that about devotional practice, in its founder's own voice, and it is the most useful thing in the chapter.
The creeper of devotion grows when watered by hearing and chanting.
And the same watering grows the weeds.1
That is the whole observation and it is worth stating flatly because traditions almost never say it. The practice is not selective. It supplies whatever is in the ground, and what is in the ground includes things the practitioner would not have chosen to feed.
So the gardener has a second job. Not more watering — discrimination. Identifying which of the things now growing is the creeper, and removing the rest, repeatedly, forever.
And the list is the part that reorganises the page, because of what is on it.
Sense gratification, obviously. Material desire, obviously.
And the desire for liberation — mokṣa-vāñchā — classed as a weed, alongside the others.2
That is not a minor entry. Liberation is the terminal goal of most of Indian religion; it is what the six philosophical systems are for; it is the thing renunciants renounce in order to obtain. And this chapter files it with sense gratification as something that grows in the same bed and has to be pulled.
Two things follow.
The corpus is not ranking goals. It is not saying liberation is a lesser attainment. A weed is not a smaller crop; it is a different plant competing for the same water, and the claim is that wanting liberation actively consumes the resource that would have grown devotion.
And it means the practice cannot distinguish them. Hearing and chanting feed both. Somebody practising devotedly while wanting liberation is growing the wrong thing with the right method, and the method will not tell them.
And there is one more thing on the list, which changes the shape of the instruction, and it is easy to miss because it arrives as a second flourish rather than a second claim.
The corpus takes those two desires — enjoyment and liberation, bhukti and mukti — and gives them a different image entirely. They are two witches, haunting the heart like ghosts.9
Which is not a decorative restatement. It is a different diagnosis of the same thing, and the two images do not agree about what you are dealing with.
A weed is out there and a ghost is in here. The weed is in the bed, it has a location, and the whole point of the horticultural frame is that you can walk over and pull it. A haunting has no location. You do not pull a ghost. You cannot get your hands on it, you find out it is present by what it does to you rather than by looking, and the standard remedy in any tradition that talks about possession is not removal by the possessed party — it is somebody else, with authority, doing something to you.
So the corpus offers a manual operation and an exorcism for one problem, and never remarks on the difference.
And the second image is the honest one. The chapter has already conceded that the two plants have identical foliage and that the distinguishing variable is invisible — which is to say the thing cannot actually be found by looking at the bed. The moment you admit the weed cannot be seen, you have stopped describing a weed. The witches are what the weed doctrine becomes once you take its own epistemics seriously, and it is the ghost image, not the plant image, that predicts where the chapter's instruction actually lands: on somebody else.
The obvious reading is that a devotional tradition is disparaging a rival goal, and there is some of that.
But look at the mechanism the corpus supplies, because it is more interesting than the polemic.
Liberation is a terminating state: arrival, completion, the end of the process. Devotion, on this corpus's account, is a relationship with no terminating condition — the summit is a rate rather than a state, the five liberations are declined precisely because accepting one would exit the process.3
So the two goals are not merely different. They are structurally incompatible: one is the desire for the thing to end, and the other is a thing that cannot end. A practitioner holding both is holding a contradiction, and the corpus's weed image is a description of what that does to a practice rather than an insult aimed at Advaitins.
That is a real argument and it survives the sectarian framing.
The chapter's practical difficulty is not stated and follows immediately from the metaphor.
Weeds and crops look alike when small. That is why weeding is a skill rather than a chore, and why it is done by people who know the plant.
And they are hardest to tell apart at exactly the stage when pulling matters most. A mature weed is obvious and has already taken the season's water; a seedling is ambiguous and cheap to remove. The window in which the operation is worth doing is the window in which it cannot be done reliably, which is a genuine feature of gardening and an exact description of the problem the chapter is posing.
And the corpus's own account makes the resemblance worse than usual. A person wanting liberation and a person wanting devotion perform exactly the same practices — they hear, they chant, they associate with devotees, they observe the same disciplines. The difference is entirely in what they want, which is the corpus's characteristic invisible variable.4
So the gardener is being asked to distinguish, in themselves, between two plants that produce identical foliage and differ only in a root nobody can see.
The corpus's answer is the association of advanced devotees — someone who knows the plant does the identifying. Which is coherent, and it relocates the discrimination outside the practitioner, and it means the tradition's most practically important operation is one you cannot perform on yourself.
That is worth stating plainly because it is the chapter's actual instruction, and it does not look like an instruction. It looks like a horticultural aside.
There is one entry on the corpus's offence lists that belongs here and the corpus does not connect it.
Chanting the holy name in order to neutralise the consequences of one's own conduct is named as the gravest of the ten offences.5
Which is a weed with a specific and awkward property: it grows only where the practice is. Nobody who has never chanted can commit it. It is not an imported desire competing for water; it is a desire that the practice itself generates, in people who have understood the practice correctly and drawn the obvious inference.
The same is true of the subtler versions the corpus lists elsewhere — chanting for reputation, for followers, for standing in a community that values chanting. Every one of those requires the practice to already be established and valued.
Which suggests a rough ordering the corpus does not supply and its own material implies. Early weeds are imported — the desires you arrived with, competing for the water. Late weeds are endemic — generated by the practice, growing in proportion to its success, and unavailable to a beginner.
That predicts something uncomfortable and probably true: the weeding problem does not diminish with progress. It changes species. A practitioner who has successfully cleared the imported weeds has, by that success, created the conditions for the endemic ones — and the endemic ones are harder, because they are indistinguishable from the crop by anybody outside and are experienced from inside as devotion.
So the bed is not simply full of pre-existing weeds. The crop produces its own, and the more successful the practice, the more of them there are — which is a description of every discipline that becomes an identity, and this corpus is unusual in having noticed.
Layer A gives the weeds, the shared watering and the list, briefly, inside the creeper metaphor.
Layer B expands the weed catalogue considerably and presses the liberation entry hard, which is the commentator's standing preoccupation.6
The shift is worth marking. The verse's weed list is a practical warning to a gardener. The commentary's is a polemical inventory of rival positions. The first is about what is growing in your bed; the second is about what other traditions want. Both are in the material and only the first is useful to a practitioner.
You have been practising something for a year and something else has grown alongside it. Assume this rather than checking for it. The practice is indiscriminate; something is always growing beside the crop; and the question is only whether you have identified it yet.
You are trying to work out whether your motivation is clean. You cannot do this alone, on this corpus's own account, and the reason is not moral weakness — it is that the two plants have identical foliage. Ask somebody who knows what the crop looks like.
Your discipline has become an identity. That is the weed that grows only where the crop is, and it cannot be prevented by more discipline, because discipline is the water. It has to be pulled, repeatedly, and it grows back.
Somebody tells you that a goal you hold is competing with the one you say you want. Before dismissing it, check whether the two are structurally compatible — whether one of them terminates. A goal that wants the process to end and a goal that has no end are not two priorities. They are one contradiction.
The evidence is a compact passage: the shared watering, the weed list with liberation on it, and the instruction to discriminate.7 It is stated as practical horticulture and its implications are considerably larger than its length.
The tension the source doesn't close: the corpus classes the desire for liberation as a weed and defends the word mukti in scripture against emendation. A convert asks to remove it from a name of God and is refused.8 The thing is a weed and its name is protected.
A second one. The discrimination the chapter requires cannot be performed by the person who needs it, because the distinguishing variable is invisible. The corpus's answer relocates it to advanced devotees, which makes the tradition's most important practical operation dependent on an external judgement — and the corpus's disciplinary chapters show how those judgements go.
Open: whether the practice-generated weeds can be removed at all, or only managed. The corpus lists them as offences and prescribes avoidance, and nothing in the range read addresses the case of a weed that regrows because the crop is healthy.
This page is the second half of The Bhakti-latā-bīja and neither works alone. That page establishes that the practitioner's only lever is supply; this one establishes that supply is indiscriminate. Together they describe a practice in which you control one variable, it feeds everything, and the sorting is somebody else's job.
Against Refusing the Five Liberations this supplies the mechanism. That page has devotees decline liberation when offered; this one explains why wanting it in the first place is a problem — not because it is beneath them but because it is a terminating goal in a system with no terminus.
And it belongs beside The Ten Nāma-aparādhas, which the vault holds from the Rūpa build. That page enumerates the offences; this one supplies the horticultural frame that makes them intelligible as a class — they are not sins, they are competing growth, and the difference changes what to do about them.
The Virtue-Signalling Spiral — that page describes a dynamic in which each display of goodness responds to the last, the baseline ratchets, and the system drifts toward positions nobody holds, because what is being competed over is the signal rather than the underlying good.
The weed that grows only where the crop is, is that spiral's origin, and the pair explains something neither page does alone.
The signalling literature usually treats the spiral as a social phenomenon: it needs an audience, competitors, and a visible scoreboard. Remove those and the dynamic stops.
This chapter locates the same thing one step earlier, inside a single practitioner, with no audience required. Chanting to neutralise your own conduct is not a signal to anybody; it is a private inference drawn correctly from a doctrine you have understood. The practice is efficacious, therefore it can be used, therefore somebody will use it — and that thought is available on day one, alone, in a room.
Which reframes the spiral's aetiology. It does not start with an audience. It starts with a practice becoming valuable, and the audience is what converts a private instrumentalisation into a public ratchet. The signalling spiral is the social phase of something that begins as a solitary weed.
And that has a practical consequence the signalling frame does not deliver. Removing the audience — no public scoreboard, no visible competition — will stop the ratchet and will not stop the weed, because the weed never needed anybody watching. The corpus's remedy is the right shape: not concealment, but somebody who knows the crop, looking at your bed.
Sharpest implication. Any practice worth doing generates its own weeds, and they grow in proportion to how well the practice is going — which means the healthiest-looking bed is the one most in need of weeding, and the practitioner is the person least able to tell the plants apart. This corpus noticed that about its own method and said so, which is rarer than the observation.
Generative questions.