Read almost any English rendering of this literature and you'll meet ghosts, spirits, demons and fiends, apparently interchangeable, apparently just atmosphere.
They aren't interchangeable. They're three different conditions with three different causes, and the difference is administrative.
Debroy sets it out in an endnote:
A preta is the spirit (not necessarily evil) of a dead person before the funeral rites have been performed. A bhuta (not necessarily evil again) is the spirit of a dead person who has had a violent death and for whom proper funeral rites have not been performed, and may not even be performed. A pishacha (necessarily evil) is often created deliberately through evil powers.1
Three states, and none of them is primarily about wickedness. Two are about paperwork.
Look at what actually separates the categories.
A preta is in a waiting state. The person died; the rites haven't happened yet. There's nothing wrong with them. The condition is temporary by design, and it ends when the living do their job.
A bhuta is a preta whose job never got done — killed suddenly, in the wrong place, with nobody to perform the rites, or nobody willing. Also not necessarily evil. Just stuck, and stuck permanently.
So the first two categories are produced by the state of a person's funeral arrangements, not by the state of their soul. A saint who dies in an ambush with no family to burn the body ends up a bhuta. A scoundrel with attentive sons doesn't.
That single fact reorganizes how the vault should read every ghost in this corpus. The frightening thing about these states isn't damnation. It's administrative abandonment.
Then there's the pishacha, and Debroy's clause about it is the one that does damage.
Necessarily evil — the only one of the three carrying that. And "often created deliberately through evil powers."
Created. By someone. On purpose.
That is not an afterlife state a person falls into. It's a manufactured condition, and manufacturing implies a manufacturer with a method and a motive.
Debroy also notes that a pishacha is "one who lives on flesh," and that in practice all three words get used synonymously — which is exactly how the distinction gets lost in translation and in reading.2
The volume's third rescue story turns on this word, and the gloss changes it.
Binduga dies, goes through hell "for a large number of years," and then — the text is careful about sequence — because he "still needed to exhaust his sins," he is born as a pishacha in the Vindhya mountains.3
Read with the endnote in hand, that sentence is doing two incompatible things at once.
The framing is penal: he hasn't finished paying, so he continues in a worse form. That's karma as a sentence being served.
But the category he's placed in is, on Debroy's gloss, one that gets made by someone using evil powers. Those are different causal stories. One says he earned it. The other says somebody did it to him.
The text never says who, or whether anyone did. It uses the word and moves on.
And nobody in the story asks. Chanchula asks where he is and how to fix it. The goddess answers with the sequence — hell, then this, and now he suffers. Nothing in the exchange treats his condition as anything other than a place he ended up.
So either the deliberate-creation sense isn't operative here and the word has softened into a general term for a nasty spirit, or the text is using a category whose implications it isn't tracking. Both readings are live. See Binduga and Salvation by Proxy.
Your father has died four days from anywhere, on a road, and the people who found him did not know his name.
You get the news eleven days later.
Here is what you now know, and it is not a metaphor to you. The window in which he was a preta — waiting, not suffering particularly, just unfinished — has closed. Nobody lit anything. Nobody said the words on the right day.
He is not being punished. He was not a bad man. That isn't the mechanism and you know it isn't.
He is stuck, because of where he happened to be standing, and because you were eleven days away.
This is why the rites matter to you in a way that has nothing to do with belief in your own piety. The obligation isn't devotional. It's closer to a debt — the dead cannot complete the transition without the living, and the living are the only ones who can.
And it is why the volume's promise about the ash lands the way it does: that a man who wears it saves a thousand predecessors and a thousand successors in his own lineage.4 That's not poetry. That's insurance against exactly this.
The practical reason is that this corpus is full of these words and the vault has no key for them.
The volume threatens people with becoming a preta, describes Binduga as a pishacha, has Shiva attended by ganas and surrounded by bhutas, and stages much of its imagery in cremation grounds. Without the distinction, all of that flattens into generic spookiness.
With it, a great deal becomes legible. The cremation ground isn't Śiva's home because it's frightening. It's the place where the transition either happens or fails — the administrative frontier — and the god who lives there is presiding over the one process that decides which of these three you become.
Line the three up by what would end them and the asymmetry is stark.
A preta ends when the rites are performed. There is a procedure, a timetable, and a responsible party. The state is designed to terminate.
A bhuta ends if the rites are somehow performed late or by proxy — and Debroy's "may not even be performed" concedes that often nothing ends it at all. The procedure exists but the responsible party is missing.
A pishacha has no such exit named anywhere. There is no rite that undoes deliberate creation. When the volume needs to release one, it doesn't perform a funeral — it sends a musician with ropes.8
So the first two are states in a process and the third is a manufactured object. You cannot un-perform a making by completing a ceremony that was never the cause.
That may be why Binduga's rescue looks so unlike everything else in the corpus. Nothing standard applies to him, so the text improvises.
The practical consequence for reading this literature is worth stating plainly, because it changes a lot of pages.
English gives us one word and it carries the wrong freight. A ghost, in the register most readers arrive with, is a person who lingers — usually because something is unfinished emotionally, and usually with intent.
None of the three works like that.
The preta isn't lingering; it hasn't been processed yet. The bhuta isn't haunting anyone on purpose; it is stranded. The pishacha isn't a person at all in the relevant sense — it is something made, and made by an agent the texts don't name.
Read "ghost" and you import a psychology. Read the three terms and you get an administrative status, a logistical failure, and a product. Not one of them is primarily about the dead person's wishes.
The evidence is a single translator's endnote, and it should be held as that. Debroy gives no source for the taxonomy and presents it as background knowledge. [SCHOLARLY — translator's endnote]
Tension one, which Debroy names himself. He sets out three precise definitions and then says "the three terms are often used synonymously."5 So he supplies a distinction and immediately warns that the texts don't observe it. The vault should treat the taxonomy as a real classical scheme that individual passages may or may not be invoking — not as a decoder ring to apply mechanically.
Tension two. The volume's own usage doesn't obviously respect the scheme. Binduga is called a pishacha and also, in the same chapter, described as ceasing to be a preta when the recitation ends: "He will no longer be a preta."6 Two categories for one entity, four sentences apart.
Tension three, unresolved and not small. If pretas and bhutas are produced by failed funeral arrangements rather than by conduct, then a person's posthumous condition is substantially determined by other people's diligence and by accidents of geography. The volume never addresses whether that's just, and it has no visible mechanism for the unclaimed dead beyond the merit-transfer described in the Binduga chapter.
Open question. Who makes pishachas? The gloss says "evil powers" and the corpus is silent. Whether this points to a real practice the texts elsewhere describe — the volume does, after all, casually list ritual counts for killing and confounding people7 — or is a vestigial etymology, cannot be settled from Volume 1.
The vault's cremation-ground cluster comes entirely from Selvalingam, and it reads these spaces symbolically. Śava/Śiva: This Body Is the Corpse holds that the only corpse the rite needs is the one you're sitting in. Śiva, Pyre to Pyre reads death as initiation.
Those are strong readings and this page doesn't contradict them. It does supply the layer underneath, which the practitioner material skips: before the burning ground is a symbol it is a facility. It is where a specific procedure is performed on a specific timetable, and the consequence of not performing it is a named, permanent, undesirable state for someone you loved.
The divergence is one of altitude rather than truth. Read only through the symbolic frame, a bhuta is a figure for unfinished business in the psyche. Read through the endnote, a bhuta is what your uncle became because the body was never found. Both are in the tradition. Only one of them explains why anybody bothered.
Alienation: The Necessary Separation and Its Terrors (psychology). The psychology corpus treats separation from the group as a primary terror — not merely unpleasant but structurally threatening, because the self is constituted in relation and exclusion attacks the constitution rather than the comfort.
The parallel here is exact and posthumous. A bhuta is precisely a being whose separation was never repaired, and the repair was always going to have to come from outside — no amount of effort by the dead accomplishes anything.
What the pairing produces that neither gives alone: this is a model of exclusion in which the excluded party has no agency whatsoever, and it clarifies what the psychology material is actually describing. The clinical literature studies people who fear exclusion and can, in principle, act to prevent it — apologize, conform, repair. Strip out that possibility entirely and you get the bhuta: the pure case, where belonging depends wholly on whether someone else does something. That's a useful limit for reading the vault's shame and belonging pages, because it isolates what remains of the terror once every self-help move has been removed.
Sacred Containers: Designing Rituals (behavioral-mechanics). The behavioral pages argue that rituals function by bounding time and attention — the container does the work, and its power is largely over the participants.
Funeral rites break that frame in a way worth noting. Here the ritual's stated beneficiary is not present and not conscious, and the entire classification of the dead turns on whether the living executed a procedure correctly and on schedule.
The insight: this is a ritual whose failure mode is assigned to a third party. The behavioral corpus consistently models ritual as something done with or to participants, where the effect is measured in them. A śrāddha's efficacy is claimed for someone who cannot attend, cannot consent, and cannot complain — which means the entire feedback loop that would normally discipline a ritual's design is absent. That's a structurally different object from a tea ceremony or an initiation, and it may explain why funerary procedure is among the most conservative and least-revised ritual material in any tradition: nothing can ever demonstrate that it didn't work.
Sharpest implication. Two of the three ghost-categories in this literature are produced by failed funeral arrangements, not by wickedness — Debroy says plainly that neither preta nor bhuta is necessarily evil. A person's posthumous state is largely decided by other people's diligence and by where they happened to die. The third category, the pishacha, is the one the volume assigns to Binduga, and it is glossed as something deliberately created by evil powers — a manufacturing claim the text uses and never examines.
Generative questions.