Four things on a table, and four people holding them.
The first holds them in one arrangement. The second holds the same four things, in the same four hands, in a different arrangement. Nothing else about them differs — not the height, the face, the clothing, the expression, or what any of the four objects is.
Are they two people or one?
Most sorting relies on some property being different — a colour, a capability, a history. Here there is no property. The only thing separating the two is the sequence, and sequence is not a quality anybody possesses. It is a fact about arrangement.
The second half of Madhya 20 spends a long time on exactly this, and the corpus does not remark on how strange it is.
The chapter's expansion scheme runs in tiers, and the tiers are real distinctions.1
Svayaṁ-rūpa — the original form. Tad-ekātma — forms of the same self, differing in feature. Āveśa — beings empowered for a function rather than being the person at all. Layered over that are the prābhava and vaibhava grades, which sort the manifestations by fullness.
That much is a taxonomy of the ordinary kind: things are distinguished by what they possess and how much of it.
Then it reaches the catur-vyūha — the four primary expansions — and the four expand to twenty-four forms.
And the sorting principle changes without announcement.
The twenty-four are distinguished from one another solely by the order in which four identical weapons are held in four identical hands.2
Not by attribute. Not by function, capacity, jurisdiction or grade. The conch, disc, club and lotus are the same four objects in every case; the hands are the same four hands. Rearrange which object is in which hand and you have a different form, with a different name, and a place in a scheme that later assigns the twelve months and the marks on a devotee's body.3
Twenty-four is what four objects in four positions yields when you keep the set fixed. The number is combinatorial, not observational — nobody counted twenty-four forms and then noticed the pattern; the pattern generates the count.
Which is worth stating plainly because it is the one genuinely interesting structural fact in a very long enumeration, and it is easy to lose inside the names. The corpus has produced a system in which identity is positional, and possession of qualities carries no distinguishing information at all.
Follow the consequence, because it cuts against how the rest of the corpus argues.
Everywhere else, this tradition sorts by degree of fullness. Kṛṣṇa over the Nārāyaṇas; sweetness over opulence; portions and reflections ranked beneath their source; the whole descent argument turning on what one form has that another lacks.4 That is a hierarchy of properties, and it is the corpus's standard machinery.
The twenty-four are not hierarchical. No permutation is fuller than another — there is no more-ness available when the contents are fixed. They are a set of equals distinguished by arrangement, sitting inside a system otherwise obsessed with ranking.
And there is a second consequence the corpus does not draw. If a name attaches to an arrangement rather than to a bearer, then the names are addresses, not descriptions. They tell you where in the scheme you are, which is exactly what an index does, and nothing about what you would encounter.
That is not a criticism of the scheme. It is what the scheme is for — and it explains why the twelve months and the tilaka marks hang off it so naturally. You do not meditate on a permutation. You locate yourself against one.
The chapter's own image for the whole architecture is unusually generous, and it undercuts half of it.
All the secondary candles are equally powerful. The original is not brighter, hotter or more capable. It is simply the one the others were lit from.5
Which is a claim about priority without superiority, and it is a genuinely difficult position to hold. Ordinarily a source outranks its derivatives because something is lost in the derivation — that is what makes a copy a copy. The candle image denies the loss and keeps the ordering, and the only thing the original retains is having been first.
Set against the permutation scheme it fits perfectly: equals distinguished by position, with the source distinguished by temporal position rather than by content.
Set against the rest of the corpus it does not fit at all. The descent argument requires the original to be fuller — sweeter, more complete, in possession of something the expansions lack, which is exactly why the majestic forms are said to defect toward the cowherd one.6 A candle that loses nothing in the lighting cannot support that.
The corpus supplies both and never asks which it means.
The compression question is unavoidable and the honest answer is not dismissal.
This enumeration — the four expansions, the twenty-four permutations, the twelve monthly presiding deities, the marks and the mantra — occupies roughly a third of the longest chapter in the book, and it yields one page. A page-per-named-form reading would have produced twenty-four pages of vault-authored filler under Kṛṣṇadāsa's name, because the forms do not differ in anything a page could be written about.
But the length is not padding, and this build has met the reason before. Ādi-līlā 10 to 12 spend three chapters cataloguing several hundred named followers, and the corpus states outright why: remembering the names is itself the practice.7 The catalogue is a sacrament rather than an appendix.
The same applies here. An enumeration you are meant to recite is not the same artefact as an enumeration you are meant to read, and compressing it to its structure — which is what this page does — preserves the argument and discards the practice. That is the correct trade for a knowledge base and it is worth saying that a trade was made.
Layer A supplies the tiers, the permutation scheme, the twenty-four names, the monthly assignments and the candle image, at length.8
Layer B does what a devotional commentary must do with an index: it converts it into instruction, attaching the marks and the mantra to daily practice and pressing the necessity of the scheme's authority.9
The direction is worth marking, and here it is benign rather than tendentious. Layer A is a map; Layer B is a set of things to do with the map. Nothing in the commentary contradicts the verse. But the sense that these twenty-four are for something — that the taxonomy is operational rather than descriptive — is considerably stronger in the purports than in the text, and a reader who takes the operational framing from the commentary will not notice that Layer A never says what any individual permutation is good for.
You are looking at a classification and cannot tell what separates two of its entries. Check whether anything about them differs, or only their position. Positional distinctions are perfectly legitimate and they behave differently: they cannot be argued about, they cannot be got wrong by observation, and they carry no prediction.
Somebody's taxonomy has a suspiciously round number in it. Ask whether it was counted or generated. Four objects in four positions gives twenty-four before anybody looks at anything, and a generated count is a fact about the generator.
You are compressing a long enumeration. Ask what the length was doing. If the list is meant to be recited rather than consulted, the structure is not a summary of it — it is a different object, and saying so is the difference between compression and dismissal.
You are told a derivative loses nothing and the original still ranks first. That is a real position and a hard one. Check whether the rest of the framework needs the original to be fuller, because most do, and a candle that loses nothing in the lighting will quietly contradict the argument the framework was built for.
The evidence is a sustained enumeration occupying roughly a third of Madhya-līlā 20, with its generating principle stated plainly enough to be assessed.10
The tension the source doesn't close: the twenty-four are equals distinguished by arrangement, inside a corpus whose every other distinction is a ranking by fullness. Two incompatible sorting principles, in one chapter, with no transition marked.
A second one. The candle image gives priority without superiority; the descent argument requires the original to possess what the expansions lack. A source that loses nothing in derivation cannot ground a hierarchy the corpus argues for continuously.
Open: what any individual permutation is for. The scheme assigns names, months and marks and never says what distinguishes one form's character, jurisdiction or approachability from another's — which is the question a devotee would actually have.
Against Sweetness Over Opulence this is the corpus running the opposite sorting principle. That page has the majestic expansions outranked and defecting toward the sweet form — a hierarchy of content. This one has twenty-four forms that cannot be ranked at all, because there is no content to rank. Both are Madhya-līlā, and the second undercuts the machinery the first depends on.
It supplies the cosmological furniture that The Three Dhāmas and the Virajā maps in space. One is a taxonomy of persons and the other a taxonomy of places, and the two together are as close as the corpus comes to a systematic ontology outside the constitutional claim.
And it belongs beside Remembering the Names Liberates, which is this build's own answer to the compression problem. That page argues that a catalogue can be a practice; this page is the case where that argument had to be applied to doctrine rather than to biography, and the same conclusion held.
The Green-Beard Effect: Kinship Hijacking Through Arbitrary Markers — that page's mechanism is a nervous system that cannot see genes and therefore reads proxies. A shared tattoo, a shared uniform, a shared marking produces the felt response reserved for kin, and the response does not weaken on learning the marker is arbitrary. The arbitrariness is not a flaw in the effect; it is irrelevant to it.
Put that beside a scheme where identity is carried entirely by the order of four objects, and the pair yields something neither has alone.
Both are systems in which a distinction with no intrinsic content does real work. A permutation is not a property; a tattoo is not a gene. In both cases nothing about the marker explains the weight placed on it, and in both cases the weight is placed anyway and functions perfectly.
But they fail differently when the arbitrariness is noticed, and that is the useful part. The green-beard effect depends on the proxy not being examined — the mechanism is a shortcut, and calling it a shortcut is a criticism of it. The permutation scheme states its own arbitrariness openly: the corpus tells you the objects are identical and the hands are identical and the order is the difference. Nothing is being smuggled.
Which gives the finding. An arbitrary marker is either a shortcut or an index, and the difference is whether the system claims the marker explains anything. The green-beard proxy claims to track kinship and does not, so exposure damages it. The permutation claims to track position and does, so exposure changes nothing. Most of the arbitrary distinctions people argue about — credentials, categories, brand marks, professional boundaries — are being defended as the first kind and would be unembarrassable as the second, and the party defending them usually does not know which they hold.
Sharpest implication. A distinction can be entirely positional and entirely functional at once, and the systems that survive scrutiny are the ones that say so. This corpus tells you outright that twenty-four forms differ only in the order of four identical objects — and the honesty is what makes the scheme unembarrassable, in a book whose other rankings all depend on somebody having more of something.
Generative questions.