The 72 Spirits of Goetia

Seventy-two entries in three ranks is one of the most looked-up lists in modern occult practice, and one of the least often read as a table. People arrive at it with a question about money or a relationship or a job interview, scan down for a name that sounds appropriate, and take the entry with the best reputation. That procedure reliably produces a poor result, because reputation inside this list has almost nothing to do with what the entries say their office is. What follows treats the material as the structured document it is. The ranks, the cabinets, the stated offices, the provenance of the underlying text, and the difference between an administrative description in a medieval compilation and a characterisation invented for entertainment four centuries later. Read as a table it becomes usable. Read as a menu of supernatural personalities it stays permanently confusing.

What the table of seventy-two is actually listing

It is a numbered inventory with three attributes attached to each entry: a rank, an office or domain, and a stated capacity. The ranks indicate status within the scheme. The office is the function claimed for that entry, expressed in the vocabulary of medieval administration. The capacity is the specific grant attributed to it, usually phrased as something the entry can provide on request. Nothing in the structure is a personality, a temperament, a gender, or a threat level, however commonly it is read that way.

That framing matters because the entries were compiled as part of a larger scheme of authority, in which lesser spirits held defined jurisdictions under a hierarchy. Read the structure and the list becomes administrative. Read the entries as characters and you get seventy-two invented personalities, because every personality detail you find in circulation was added by a later writer with no access to the source.

The three ranks and the uneven numbers

The counts are nine Presidents, thirty-six Dukes and thirty Kings, which totals seventy-five rather than seventy-two and has caused confusion for as long as the list has circulated. The discrepancy is a feature of how the material was assembled rather than an error to be corrected by anyone in particular. Different manuscripts and different printed editions handle the arithmetic differently, and modern retellings copy whichever number they happen to encounter first.

If you are reading the list for structural understanding rather than for a working, the exact count is the least interesting thing about it. What carries the meaning is the tier structure: a small top group associated with broader authority, a large middle group associated with specific trades and functions, and a large lower group associated with the physical and remedial. It is a hierarchy, and the tiers tell you how much latitude the entry is supposed to have.

The five cabinets and what grouping does

The entries also fall into five groups, conventionally labelled as cabinets, which cut across the rank divisions. Grouping them this way is unusual and it is not arbitrary: each cabinet gathers entries whose stated functions share an area, so the grouping is a functional index rather than an organisational chart. It is genuinely useful, because it lets you narrow a long list to a handful of candidates without reading all seventy-two descriptions.

It also has a practical weakness. A question that spans two areas, which is most real questions, will not sit neatly inside one office, forces a choice that the table itself does not make necessary. Where that happens, the better move is to select from two cabinets and run two short separate workings rather than forcing a single entry to do a job it was never described as doing.

The nine Presidents, by office rather than reputation

Nine entries occupy the top rank. In the standard ordering they lead with Bael, then Agares, Samigina, Marbas, Valefor, Aamon, Belphegor, Saleos and Vassago. Their stated domains cover rulership, teaching, discovery, crafts, investigation and fortune. This is a fairly narrow and rather workmanlike set of jurisdictions, and it reflects an administrative scheme rather than a dramatic one. Several of the descriptions are closer to a magistrate or a master of a guild than to a sovereign.

Bael and Marbas and Vassago are probably the most often named of the nine because their domains are legible at a glance. That is a reason to be cautious with them rather than a reason to start there. The entries whose function resists a one-line description are often the ones whose descriptions are most specific, and specificity is what makes an entry useful for a defined task rather than for a general wish.

The Dukes, and where the wealth material sits

Thirty-six entries occupy the middle rank and carry the most usable descriptions in the whole document, because a duke in this scheme is a functionary with a defined brief. The wealth-related entries are the most worked with and the most often mishandled. Clauneck is described as granting riches and is a duke, not a top-rank entry. Mammon is associated with wealth and holds a lower position. Paimon is associated with dignities and rank. These three are constantly swapped for one another in online material, usually with no reference to what the source actually says each one does.

The office is the thing to select on. If your question is about money specifically, an entry whose stated capacity concerns riches is a better match than a famous one whose capacity concerns something else. If your question is about position, an entry about dignities is the correct starting point and a riches entry is not. Confusing these two is the most common selection error in the list and it produces a working aimed at the wrong target entirely.

The thirty Kings, and what kingship signifies

The lower rank holds thirty entries described as kings, which is counter-intuitive and worth understanding. Within the source's own logic, the term appears to be applied to entries granted authority over particular spheres rather than over a general population, and the titles overlap with the dukedoms. Several of the descriptions in this rank concern animals, objects, positions and natural phenomena rather than human affairs.

The practical consequence is that this rank is under-used and sometimes misread as the important one because of the word. It is not. The names in it are frequently striking, and striking names attract attention, but the offices described are narrow. Selecting on the basis of how memorable a name is rather than what it governs is the mechanism behind a great deal of unproductive work in this corpus.

Reading an entry as an office rather than a personality

Take the description at face value and strip everything that was added later. What remains should be a function: teaches, provides, discovers, governs, restores, kills, destroys, rewards, inflames, causes invisibility. These verbs are the actual content of the list. An entry that governs horses is a horse entry, and its reputation in film or games has no bearing on what the text attributes to it.

A useful discipline is to read each entry twice: once for the stated function, and once for the conditions attached to it. The conditions are almost always where the useful information sits, because they specify what must be offered, what must not be asked, and what happens if the terms are ignored. The conditions were written by a compiler attempting to regulate a dangerous procedure. The functions were written by the same compiler, but the conditions are where the regulation shows through.

Why fame is the worst selector in the list

Reputation inside this corpus is a product of four centuries of retelling, and the most visible entries are prominent on grounds unrelated to their suitability. A name that is short and hard to pronounce travels. A name attached to a memorable story travels. A name attached to a film travels fastest of all, since the film needed a figure that general audiences would recognise rather than one that matched the text.

The practical cost of selecting this way is that you will repeatedly choose an entry whose stated domain does not cover your question, and then attempt to stretch its description until it fits. The stretch is invisible in the log, because the log records your question and not the mismatch you ignored. If your practice is not working, an audit of past selections for office mismatch is more informative than any change to the working itself.

The two entries most often chosen for the wrong reason

Two selections recur constantly and are almost always made on grounds unrelated to the description. The first is a top-rank entry chosen because its name is well known, when the question at hand concerns a narrow domestic or commercial matter that a mid-rank functional entry was written for. The second is an entry chosen because it is described in a dramatic way in retellings, when the original description is bureaucratic and the drama was added to make the material more sellable.

Neither error is fatal, and both are easy to fix retrospectively. Note the office, note the question, and check whether they overlap. Where they do not, the working has been aimed at something other than what you asked for, which is a sufficient explanation for a null result that does not require any conclusion about magic.

A selection sheet to fill in before you begin

Before choosing an entry, write four lines. The question, in one sentence and specific enough to be answerable. The office required, in the vocabulary of the list. The cabinet it falls under. The rank and capacity as stated in the source, quoted rather than paraphrased. Only then select, and record the reason for the selection in a fifth line.

The sheet takes five minutes and it converts a vague question into a specification. Most of the value comes from the second line, because naming the required office in the list's own vocabulary usually identifies two or three candidates immediately, and the remaining judgement is a small one rather than a large one. A list of seventy-two becomes a shortlist, and the shortlist is defensible.

1. The question, written so that an answer could be observed.
2. The office required, using the vocabulary of the list itself.
3. The cabinet the office sits in, and the rank of each candidate.
4. The stated capacity, quoted exactly rather than remembered.
5. The reason this entry was chosen, written before the working.

Provenance, and how far the material actually goes back

These seventy-two names enter the European grimoire lineage through two older channels: a seventeenth-century conjuration text, and the Lemuria. Much of the name material traces back further still, to a category of angelic being recorded in Jewish and Christian demonological writings including the Pseudepigrapha Book of Enoch. The compilation as an inventory of seventy-two specific entities with ranks and offices is a comparatively late arrangement, but the names themselves are considerably older.

This matters for two reasons. First, the list is a synthesis rather than a survival, so its internal structure tells you about the compiler and the period rather than about whatever the names originally referred to. Second, the modern branded form, which is dominated by video essays, apps and commentary, is a twenty-first-century layer sitting on top of all of it. There is a modern product here as well as an old text, and confusing the two is the root of most of the strangeness people encounter.

What the classical context changes

In the source's own scheme these were administrative or disciplinary officials rather than tormentors of the innocent, and they operated within a hierarchy under a greater authority. That framing was largely lost when the material was removed from its context and circulated as a catalogue of evil personalities. The change is not cosmetic, because it inverts what a reasonable person should do: an official with a brief is a matter of procedure and terms, while a tormentor is a matter of threat and avoidance.

Working from the administrative framing produces a more constrained and considerably more defensible practice, because it makes the conditions in each description feel like the point rather than as a footnote. It also removes the escalation that drives people into worse deals: if the entity is an official, then a request that demands escalating payment is a request being refused, and the appropriate response is to withdraw rather than to escalate.

A ranking sheet that does not require an opinion

If you want to compare candidates without a personality contest, rank them on three measurable things instead. How closely the stated capacity matches your specific question. How many conditions and restrictions the entry carries, since a heavily conditioned entry is a more documented one and gives you more to work with. How well the entry's rank suits the scope of the request, on the reasoning that a broad request and a narrow function are a poor match at any tier.

Rank those three columns and the winner is usually not the famous one. This is an unfashionable result and it is the reason a table-based approach is worth the effort of building. Every element of the method is drawn from the source text, so a selection can be audited by someone who does not share your commitments, and that auditability is the thing that separates this approach from selection by vibe.

Which entries to leave alone at first

Some entries carry severe conditions attached, and the source itself attaches them for a reason. Entries described as destructive, entries associated with loss, and entries whose conditions require ongoing obligations are poor first choices even when the office matches, because you will have no framework for interpreting what a difficult outcome means and no way to tell a failed working from a misunderstanding of terms.

Begin instead with an entry whose capacity is straightforward, whose conditions are light, and whose domain you can verify in the log. Almost all competent practice in this corpus starts with something undramatic, and the people who begin with the dramatic entries are the ones who end up with a practice built on results they cannot interpret. Learn the structure on a small question and the larger ones become available to you as a choice rather than as a test.

Frequently Asked Questions

Is a paid spirit browser worth it if I already have the printed list?

The print list is the source, so nothing is missing from it. What a browser adds is search, filtering and portability, and that is worth something if you intend to work through the list systematically rather than looking things up occasionally. What it does not add is interpretation, and the interpretations are the part where errors are most likely to enter. If you buy one, the discipline that makes it useful is writing your selection reason in your own log before you begin, not after.

Which of the nine Presidents should someone start with?

None, on the basis of the text alone, because the top rank is described in broad terms and a broad term is a poor match for a specific first question. Pick an entry from a middle rank whose stated capacity names your actual concern, and read the conditions attached to it carefully before you start. If you have a particular reason to want a top-rank entry, write that reason down, because the reason will tell you more about the working than the choice of entry will.

Are any of these actually dangerous to work with?

Nothing in the material supports the idea that the risk is supernatural in origin. The documented risks are ordinary and well understood: a working built on externalised desire can produce financial impulsiveness, a working framed as a bargain can normalise escalating obligations, and a working that obscures your own agency is bad for you independently of what any text claims. Treat the conditions as the substantive part and the risk profile is manageable.

Does rank tell me how powerful an entry is?

It tells you where the entry sits in a hierarchy described by a compiler, which is not the same as describing capacity. The source's logic appears to grant broader authority to the top rank and narrower briefs to the lower ranks, and reading that as a power ranking is a fairly modern import. For practical purposes the more useful question is whether the entry's brief covers the request, which is a better predictor of a useful working than its tier.

Should I use the modern brand material or the older text?

Use the older text for the structure and the modern material for the formatting, and be aware that you are mixing layers when you do. The offices, ranks and conditions come from the compilation. The personalities, the dramatic framings and most of the ordering that people recognise come from later retellings. A log built on the offices will survive contact with a skeptic; a log built on the personalities will not, and that difference matters if you ever want to review your own record.