Entity Lore and Invocation

What follows is an account of a working that is almost certainly a mistake, made deliberately, because the two systems it borrows from are constantly conflated and the confusion produces specific and predictable failures. One system is the entity catalogue, which supplies a vocabulary of role, form, office, and power. The other is the servitor, which is a mechanism: an attention bound to a task, with a trigger, a place, and a term. They are not variants of one practice. The catalogue gives you words. The mechanism gives you a job. Nearly everything that goes wrong in workings of this kind comes from borrowing the vocabulary and then assuming the mechanism came with it, and the symptom is a structure that looks like an entity working, gets charged like a sigil, and produces nothing attributable to anything. I built one of these, for producing dreams, and kept careful notes on where the two systems pulled apart. The notes are more useful than the working was.

A servant borrowed from a catalogue

The working was for dreams. Specifically: not for having interesting dreams, which I could not ask for, but for remembering them, which is a thing I could actually do. That distinction was the first useful decision in the whole project and I made it in about four minutes after six weeks of being unclear about it.

Having decided, I went looking for a source, and here is where the two systems got mixed, because a servant needs a form and the catalogue is where forms are written down. What the catalogue offers is a vocabulary: a name, a rank, a visible appearance, an office, and a list of powers. What the mechanism needs is none of that. It needs a task, a trigger, a location, and an end date. I had five columns of the wrong kind of information and four of the right kind, and I did not notice for a month because the form made the whole thing look finished.

The two systems being mixed here

Classical ceremonial practice has a two-term distinction that modern usage routinely collapses. Evocation is calling something to appear. Invocation is inviting an influence to operate through the operator. The first is about a presence, the second is about your own conduct, and the second is far closer to what the mechanism requires.

The catalogue is the evocation side. It is the most structured entity material in Western grimoire literature, giving each of seventy-two entries a rank, a form, an office, and specific powers, and its value is real: it is a filing system, and a filing system is genuinely useful for finding material you did not know you were looking for. What it is not is a mechanism. Nothing in that structure makes a task happen. It is a description, and descriptions do not run errands.

What the catalogue supplies and what it does not

It supplies four things, all of them linguistic. A name, which makes the servant a referent rather than a mood. A role, which gives a sentence to encode. A form, which gives a shape to hold in attention. And a set of powers, which are the least useful of the four and the most quoted, because they are borrowed descriptive language about a category rather than an instruction to anybody.

What it does not supply is the part that does the work. A task with a countable definition. A trigger, which is the specific and unromantic moment at which the attention engages, and which is the whole mechanism compressed into a single condition. A location. A dismissal with a date on it. All four are unglamorous, and all four are what the tradition gives you when it is talking about a servant rather than about a spirit, and I had read considerably more about the second than the first.

A role is not a form

The entry I started from had a specific appearance, and I reproduced it, and it took me a long time to notice that reproducing an appearance is a decision about what the thing is. If the servant is a bound piece of attention, then drawing a face for it is a way of making a memory more vivid, which is fine and is not the same as giving it a body. If the servant is a described entity, then the appearance is a description and reproducing it adds nothing.

The distinction has a practical consequence I did not anticipate. Once I had drawn a face, I found myself protecting it, which is a commitment the mechanism does not require and which made a two-minute task feel like something I was responsible for. The drawing had recruited a relationship. Nothing in the trigger or the task needed a relationship, and having one made the retirement harder than it should have been. This is the cost of mixing systems, and it is paid in obligation rather than in disappointment.

Why the borrowed name does less work than it looks

A name from a catalogue is a piece of shared reference, and shared reference is doing something real. It gives you a compact handle. It puts you in contact with a body of material and with other people who have used the same entry. When you write the name, you are pointing at a large amount of accumulated description rather than at your own intention.

That is exactly the problem. The name is doing the remembering, not you. I built a servant, charged it, and then found that when I checked in at the trigger I was checking whether the servant was in a good mood, which is a form of asking a fictional entity about its own state. The role and the form together were answering the question I should have been answering, which is whether the trigger had occurred. Nothing in the structure prevents this, and nothing in the structure encourages it either. It is a habit the vocabulary forms.

The task, which is the whole mechanism

A trigger is a conditional, and the conditional does all the work. At this hour, having slept, on waking, in this state, do this small thing. That is the entire technique, and it is a technique that any behavioural work in any field would recognise, and the tradition has been describing it for a century under a different name.

This is the part I would hand to somebody starting. Whatever else you take from this account, take the idea that the mechanism is a conditional rather than a presence. When I finally wrote mine as a conditional, the working changed character overnight, because a conditional can be checked and a presence cannot. I now write the aim as a sentence with an if in it, and I have found that the if-construction eliminates most of the inflation that the catalogue vocabulary invites, without requiring any argument about what the servant is.

A dream is not an outcome

The obvious error, which I made in the first draft and which is worth naming because the mistake generalises. I wrote an aim about dreams, plural, unspecified. A servant cannot be given that, because there is no way to tell whether it was answered, and a working whose success cannot be told apart from ordinary variation is a working that teaches nothing.

The correction was to aim at recall rather than production. At waking, before anything else, three lines: where I thought I was, what happened, and how I knew I was dreaming. Now the aim is countable, the record is countable, and the working can fail visibly. The dreams themselves remained entirely up to me, which is correct, since they are not mine to produce. This is the same correction that applies to almost every aim in this field, and the fact that it is available in a servant working makes it easier to see than in a sigil working, where a vague aim can be encoded very happily.

Designing for recall rather than production

Once the aim was three lines at waking, the rest of the design followed and it followed easily, which is the test of a good design. The trigger became the waking moment. The task became writing, which takes under a minute. The count became a month. The location became the notebook by the bed, which is a real location and does real work, because a servant stored nowhere is a servant in no particular relation to your life.

What I had was a two-minute nightly practice that produced a countable record, and the record showed something I could check, which is that recall went up and dream content did not change. That is a small result with an obvious alternative explanation, which is that writing three lines makes you attend to dreams and attending improves recall. I have no way to separate the two and I do not think there is a way to separate them, and the servant framing made me expect a separation that was never available.

Where the two systems pull apart

Three places, and they are the places where you will feel the seam. The first is the question of address: a catalogue entry is addressed, and a servitor is triggered. Addressing a thing produces a relationship and a request. Triggering a condition produces a behaviour. The second is the question of consequence: a described entity may be disobeying, refusing, or requiring something, and a servitor cannot, because it has no will to disobey with.

The third is the question of duration, and this is where I got the real cost. A catalogue entry has no end. It is on the page, available indefinitely, and there is a strong pull toward keeping the arrangement open. A servitor needs a term, and writing the term down is the step that the vocabulary makes feel unnecessary, because nothing in a description of a spirit suggests that it expires. My first working had no end date and I retired it four months later by an act of will that should have been a scheduled step.

Choosing between them honestly

If you want a name, a role, a form, and the pleasure of a piece of shared reference, take the catalogue. Use it as a filing system, take the role as your sentence, and accept that you are doing something symbolic rather than mechanical. That is a perfectly good thing to do and the tradition has plenty of material for it. Just do not then go looking for a mechanism, because the catalogue is not offering one and the absence is not a defect in your working.

If you want something that changes your behaviour, take the servitor, and take the four unglamorous fields. Task, trigger, location, term. Write the term down. And accept the cost, which is that a mechanism is a commitment rather than a relationship, and a commitment has to be maintained and eventually ended, and nobody finds that satisfying. Most workings that go wrong do so by wanting a relationship out of a mechanism. Pick one and pay its price.

The paperwork, kept short

Six lines, which is what this material actually needs and what nobody writes. What the servant is for, in one sentence with a number. When it engages, in one sentence with a time. Where it lives. What it is not for, which is the line that prevents drift and that I have never yet seen written down in somebody else's account of this kind of working. When it ends, with a date. And one line recording what happened, kept in the same place as the record.

The fourth line is the one I would emphasise. Almost every drift I have read about in this field, and several I have caused, is a servant that has quietly acquired a second and third job. A servitor that reminds you to drink water is fine. A servitor that also reminds you about money and about an unresolved situation is three servants wearing one name, and the attention has not multiplied, so what has happened is that all three became less reliable. Writing the not-for line at the start is the cheapest available insurance.

The row worth using

Not a row from the catalogue. A row of your own, with the five fields filled in, dated, and filed. The catalogue is worth consulting exactly once, for the vocabulary, and what you take from that consultation should be a single word. Everything else in the entry is borrowed language about a category, and borrowed language about a category is the most expensive thing you can put in a mechanism, because it arrives with a whole set of unexamined assumptions about what such a thing is and what it owes you.

The working I ended up with is one sentence, one conditional, one location, one month, and one date at the end of the month on which it stops. The catalogue is on the shelf, and the only thing I took from it is the word for the role, which is in the sentence. I would put that row on a single card and be able to hand it to somebody else, and the handing over is the test: if the card cannot be handed over, it has a face drawn on it somewhere, and the face is doing work that the mechanism does not need.

Frequently Asked Questions

What is the difference between a servitor and an entity?

A servitor is a mechanism: an attention bound to a task, with a trigger, a location, and a term. An entity is a described figure with a name, a form, an office, and a rank. The entity gives you vocabulary, which is genuinely useful, and the servitor gives you a conditional, which is what actually changes behaviour. The common failure is borrowing the entity's vocabulary and then expecting the servitor's mechanism, and the symptom is a working that looks complete and produces nothing attributable.

Do I need to draw a form for a servitor?

Not for any mechanical reason, and it has a cost I did not expect. Drawing a face made me feel responsible for the thing, which recruited a relationship the mechanism does not require, and that made the retirement much harder. A role, a colour, or a plain word does the same work. If you draw a form, treat it as making a memory more vivid rather than as giving something a body, and write the term down at the start, because a thing you have drawn feels like a thing you have made responsible for.

How do I decide whether a working actually did anything?

Make the aim countable before you charge it, and count something inside your own control. In this case the aim was three lines written at waking, so the record is a page of entries rather than a judgment about whether the dreams were better. That control does not separate the working from the ordinary effect of paying attention, and I do not think any of this material offers a way to do that. What a countable aim does is make a failure visible, which is worth a great deal more than a result you cannot rule out.

How long should a servitor last?

Long enough to produce a countable record and short enough that you will not resent it. Mine ran a month, which was long enough for the recall rate to move and short enough to retire without drama. The term has to be written at the start, because a catalogue description has no end and creates a pull toward keeping the arrangement open indefinitely, while a mechanism with no term is a piece of attention with no exit. The scheduling of the end is what makes it a commitment rather than a relationship.

Is it wrong to use entity vocabulary for a servitor?

No, and it is the normal way round in practice. The vocabulary is public, well developed, and gives you a role to encode, which is genuinely the hardest sentence to write. The mistake is not using the vocabulary. The mistake is assuming the vocabulary brought a mechanism with it, and then describing the result in the vocabulary's terms, which is a way of never being able to say the working failed. Use the words, keep the four unglamorous fields, and the mixing becomes a technique rather than a problem.