Three people in the same temple courtyard at seven in the morning. All three are doing what looks like the identical thing.
The first has brought fruit and flowers and is offering them with real care, and he is offering them because his daughter's visa decision comes through this month. He would tell you honestly, if you asked, that he is asking. The offering is beautiful and it is a request.
The second isn't offering anything. She's sweeping the far corner, and she has been kind to everyone who came in, and she is honest, and she eats little, and she doesn't gossip. If you listed her virtues you'd run out of paper. She is, by any ordinary measure, the best person here.
The third is doing something small and slightly boring. She is reciting names, out loud, and she is not thinking about an outcome, and she is not visibly virtuous in any way you could point to.
This tradition says all three are doing bhakti — and puts them in three different categories, in that order, worst to best. And the woman with the paperless list of virtues is in the middle.
That ranking is the content of this page, and it is stranger than it first looks.
Viśvanātha gives the division early, in Text 1, as something you should also remember: bhakti is of three types.1
Āropa-siddhā — activities which, although not consisting of pure devotion, are designated as devotion because they're offered to the Lord. Āropa means placing-upon, imposition, attribution. Devotion has been laid over them from outside.
Saṅga-siddhā — endeavours associated with or favourable to devotion, but not in and of themselves composed of it. Saṅga means contact, company, association. They count by keeping the right company.
Svarūpa-siddhā — endeavours purely constituted of devotion. Svarūpa means own-form. They count by being it.
Three ways of belonging to a category: by attribution, by association, by constitution. Only the third is the thing itself.
The first grade is the easiest to see and the least comfortable to think about, because most religion, most of the time, is here.
The book is precise about the mechanism. A person knows that without devotion's assistance, action yields no fruit. Knowing this, he dedicates his prescribed duties to the Lord's satisfaction so that the Lord might fulfil his extraneous desires.2
So the structure is: I have a want. I know the machinery requires his cooperation. I offer him my activity, sincerely, so that the machinery turns.
The book gives this two names when it's mixed with desire for enjoyment — sakāma-bhakti, devotion with desire, and saguṇa-bhakti, devotion with quality. And it does not throw it out. Because they offer the fruit of their activity to the Lord, it is considered to be a type of bhakti. Attributed, real, counted.
But notice what has happened to the direction of the sentence. His pleasure is in the middle of the transaction rather than at the end of it. He's not the point; he's the mechanism by which the point is reached. The offering is genuine and the arrow goes the other way.
The second grade is where the surprise lives, and it takes a specific text to deliver it.
The book quotes Prabuddha Muni's instruction to Mahārāja Nimi — twenty-six items of bhāgavata-dharma. Read them:3
compassion for others, friendliness, offering respect, cleanliness, austerity, tolerance, silence, study of the Vedas, simplicity, celibacy, non-violence, equanimity in heat and cold and happiness and distress, perceiving the Lord everywhere, living in a secluded place, renouncing family attachments, remaining satisfied with gain that comes of its own accord
That is a serious ethical program. It is most of what most traditions mean by a good life. And Viśvanātha's ruling on it is flat: these are not by nature purely constituted of devotion.
They're parikaras — associates, retinue, the people who travel with the king and are not the king. They acquire a likeness to devotion by being established as its assistants.
Then the sentence that does the real damage:
If bhagavad-bhakti is removed from the twenty-six qualities, then Bhagavān has no direct relationship with them.
Compassion, honesty, non-violence, self-restraint — detached from devotion, God has no direct relationship with them. They're good. They're not this. Their entire devotional standing comes from what they happen to be standing next to.
The third grade is described almost tersely by comparison, and the terseness is itself informative.
Svarūpa-siddha-bhakti is all favourable endeavour — hearing, chanting, remembering and the rest — plus the sentiments that appear from the stage of bhāva onward, provided they're free of desires separate from Kṛṣṇa and uncovered by knowledge and reward-seeking action. Put more simply: all endeavours of body, mind and words related to Kṛṣṇa and performed exclusively and directly for his pleasure, without any intervention.4
That last phrase is the whole definition. Without any intervention. Nothing sits in between the act and its object. No outcome the act is routed through. No virtue the act accrues on the way.
Which is why the list is so unimpressive. Reciting names. Listening to stories. Recalling something he did. These are not achievements. They have no by-product you could show anyone. The mark of the highest grade is that the act has nowhere else to go.
And then comes the ruling that makes the hierarchy operational rather than descriptive. In the conversation between Caitanya and Rāya Rāmānanda, both āropa-siddha and saṅga-siddha devotion are described as external.5
Not false. Not forbidden. Not sinful. Bāhya — outside.
I keep coming back to how much work that one word saves. It lets the tradition hold two things at once that would otherwise tear: everything the courtyard is doing has real devotional standing, and two-thirds of it is not the thing. No one is being told they're wasting their time. They're being told where they are.
You've kept the practice for eleven months and you're aware, in a way you'd rather not be, that you started it in a bad month and it worked. Things got better. You've never quite untangled whether you're still here for him or for the way the mornings feel now.
Under this scheme that's a live question with an answer, and the answer isn't a verdict on your character. It's a location. Some of what you're doing is grade three and some of it is grade one, and they're mixed in the same forty minutes.
Now watch what happens when you try to fix it by being better.
You get stricter. You eat less. You stop the small dishonesties. You're kinder to the person at work you'd written off. All of it real, none of it easy — and every item on that list is grade two. You have improved substantially and moved sideways. The improvement is not the thing; it's the retinue.
Third, the one that actually shifts something. Ten minutes on a Tuesday, and you're reciting, and for a stretch of it there's no reason. Not virtue, not request, not the pleasant feeling. Nothing routed through anything.
You won't notice at the time — that's the nature of it, there's no by-product to notice. You'll find it afterwards, as an absence: a stretch of the morning with nothing to show for itself.
Grades one and three are easy to defend. Grade two is where I think the real claim lives, because it's the one that costs something.
Putting the twenty-six virtues in the middle means the tradition has declined the obvious and generous move — declaring that goodness is devotion in another key, that the compassionate atheist is a devotee under a different description. That move is available, it's attractive, and plenty of traditions have made it.
This one doesn't. It says: compassion is an assistant. Non-violence is an assistant. Their standing is derived, and remove devotion and God has no direct relationship with them.
Two things follow, pulling opposite ways.
Uncomfortably: a system that ranks reciting names above non-violence has made itself available for abuse, and the history of devotional movements shows the abuse taking exactly the predicted shape — ritual precision above ordinary decency, with a doctrinal warrant for the ordering.
But there's a harder-headed reading too. The claim is not that virtue doesn't matter. It's that virtue is not self-certifying as devotion — and that keeps a door shut that most religion leaves open. If being good counted as devotion, then devotion would be indistinguishable from respectability, and the practice would collapse into whatever the surrounding culture already approves of. Grade two exists to stop the tradition congratulating its most socially admirable members on having arrived.
Which is a real cost paid for a real protection. The book doesn't weigh them. It just files the virtues and moves on.
One structural note, because it explains the shape.
You could get most of this with two categories — is it devotion or isn't it. Attribution and association would both go in the second bin.
The tradition uses three, and the reason is that the two lower grades fail differently. Grade one has the right object and the wrong direction: it aims at him, and through him at something else. Grade two has the right direction and no object: it aims at nothing in particular, and gets its standing by proximity.
Wrong direction, absent object. Two distinct diagnoses, two distinct repairs. Collapse them into one bin and you lose the ability to tell the man with the visa problem from the woman with the broom — and they need different things said to them.
Strongest support. The division is scriptural and cited: the twenty-six qualities come from a named Bhāgavatam passage, and the external ruling comes from the Rāmānanda conversation. The three terms are technical and their etymologies do the sorting work — attribution, association, constitution — so the hierarchy isn't asserted, it's built into the names.
The tension the book does not close. Grade two is stated and then dropped. If the virtues have no direct relationship with God once devotion is removed, what exactly is a devotee's obligation to them? The book never says they may be neglected. It also never says they may not, and it spends no words on the case of the ritually exact and personally cruel practitioner — who is, on this taxonomy, correctly sorted above the compassionate outsider.
A second tension. Grade one includes people offering their duties so that God will grant extraneous desires, and this counts as bhakti. But Text 1's own definition requires being empty of desires other than his pleasure. So āropa-siddha-bhakti is called bhakti by a book that has just defined bhakti in terms that exclude it. The resolution has to be that uttama-bhakti and bhakti are different words with different extensions — which is coherent, and which the text never spells out, and which means the word carries two meanings within a single chapter.
Open question. Does the middle grade have an exit? The virtues become devotional by association with devotion. Does sustained practice ever convert them into svarūpa-siddha, or do compassion and honesty remain permanently retinue, no matter how far the practitioner goes?
Read Bryant's account of the nine practices and you come away with devotion as a set of activities you take up — hearing, chanting, remembering, serving — a repertoire. It reads as additive. You do these things, and you are doing bhakti.
Come at it through this taxonomy and the repertoire stops being the point. The same nine activities can be grade one or grade three depending on what they're routed through, and no inspection of the activity will tell you which. What's being sorted isn't behaviour at all.
That's the second time this book has done exactly this move — Text 1 moved the criterion from effect to disposition, and here it moves it from activity to what the activity passes through on its way out. Same instinct twice: whatever is visible is not the variable.
And the convergence with the wider bhāgavata-dharma material is worth naming, because it's a genuine disagreement inside the same corpus. That twenty-six-item list is presented in its home context as the substance of devotional life. Quoted here, it becomes the retinue. The list didn't change. Its rank did, and this book doesn't flag that it has demoted its own scripture's language.
A category with three grades of membership — constitutive, adjacent, attributed — is the same structure as an honest signal, an enabling signal, and a mimicked one. Set this beside Signal Mimicry Arms Race — Scam vs Legitimate Branding. That page's whole subject is that a signal's meaning depends on whether it is produced by the thing it indicates, merely correlated with it, or manufactured to resemble it — and that the three are often indistinguishable at the surface. Svarūpa-siddha is the signal produced by the thing. Saṅga-siddha is the correlate travelling alongside it. Āropa-siddha is the form without the substance behind it.
What the pair produces that neither gives alone: both systems are forced into three grades rather than two by the same discovery — that surface inspection is useless, so membership has to be defined by provenance instead. And both then hit the identical consequence: once you sort by provenance, the honest and the mimicked look the same to any observer, including sometimes the producer. Signalling research answers this with cost — the signal that's expensive to fake is trustworthy. Bhakti has no such lever, because grade one and grade three cost the same to perform: reciting names is cheap whoever is doing it, and offering fruit is cheap whoever is asking. Which is why this tradition ends up placing its trust in a person rather than in a test — the qualified teacher, again, doing the work that costly signalling does elsewhere. Where you can't price the signal, you appoint a reader.
Sharpest implication. The middle grade is where every tradition's honesty gets tested, because it's where you have to decide whether being good counts as being in. Say yes and your practice dissolves into local respectability. Say no and you've built a warrant for the ritually precise and personally awful. This book says no, states it once, and does not look at what it costs.
Generative questions.