The Epithet Trap

Status: ● full (2026-07-30) · Badge: method

Related: neb-pehti · ur-heka · kherty · anit · buchis · wepwawet · khemi · the Door of the Hall of Two Truths · the Names of Osiris

Why this page exists

This reference began from an aggregate list of names said to be Egyptian deities, and set about checking each one against primary texts. A large minority of them are not deities.

They are not fabrications either, which is the interesting part. Almost every one is a real Egyptian word, correctly transcribed, taken from a real text — and then filed in the wrong category. A title became a person. A door became a goddess. A verb became a god.

Seven distinct ways this happens have turned up so far, each found the hard way and each with a worked example. They are set out here because the pattern is not confined to this reference: any list of Egyptian gods assembled by harvesting names from translations will contain all seven, and a reader who knows the shapes can spot them.

1. A title, read as a member of the roster

The shape: a text lists a group of gods and gives the group a leader, or a collective title. The title is transcribed like a name and joins the list as a thirteenth member.

Egyptian makes this easy because divine titles look exactly like divine names — Neb- ("lord of"), Khenti- ("foremost of"), Ur- ("great of") begin both.

Why it is hard to catch: the title is genuinely in the text, in the right place, spelled correctly. Nothing about the string is wrong. Only the surrounding sentence tells you it is a description of the group rather than one of its members.

2. The first element of a compound, indexed as a whole name

The shape: a god or an object has a long compound name. Something — a scribe, an indexer, a scraper — takes the first element or two and treats it as the name.

Worked example: neb-pehti. The census carried her as "a goddess." She is the left leaf of a door. Her full name is Neb-pehti-thesu-menmenet, and it is the name of one half of the doorway of the Hall of Two Truths — the other leaf being Neb-Maat-heri-tep-retui-f. Four occurrences across three books, all of them that door, and no bare Neb-pehti anywhere.

Why it is hard to catch: truncation produces a plausible name. Neb-pehti means something like "lord of strength," which is a perfectly good divine epithet, and there is no internal evidence that anything is missing.

3. An element that is used both ways

The shape: the same word is a common noun and part of divine names, and the two uses sit side by side in the same book.

Worked example: qerert. It means a cavern or circle, and Budge lists it as one of four ordinary words for a division of the underworld — "Field," "City," "Hall," "Circle." It is also inside the name of the Eleventh Division (Re-en-qerert-apt-khatu), inside the title "lord of Qerert" applied to Horus, and inside qerti, an assessor whose name is its dual — the two caverns, glossed by Budge as the two sources of the Nile.

Why it is hard to catch: every one of those uses is legitimate. You cannot rule the word out; you have to read each occurrence and decide which use it is.

4. A variant that collides with a different word

The shape: searching for a name requires generating spelling variants, because the nineteenth-century transliterations do not match modern ones. Some of those generated variants are real words that mean something else, and they return hits.

Worked examples, three of them, all mappings this reference invented and then withdrew:

kherty → Khenti. But Khenti means "foremost" and occurs about ninety times

as an ordinary word.

djefa → tchefau. But tchefau is a class of food — "thy ka with hu and

tchefau foods."

daua → the token in the archive is a verb: "daua — magnifieth him."

And the correction to the correction. Withdrawing a bad mapping is not the same as concluding the name is empty. Two of these were re-examined and the god turned out to be real under a different spelling:

kherty — Khenti was wrong, but Kherti, with the r,

is attested: "He has freed king Teti from Kherti, he has not given him to Osiris."

buchis — the mapping to Bakhau was withdrawn because

Bakhau is the Mountain of the Sunrise, a place. Correct. And Bakha is the bull, and the bull is titled "the bull of the Mountain of the Sunrise (Bakhau)." The withdrawal was right and the hunch was not wrong.

Why it is hard to catch: the search succeeds. Hits come back. Nothing signals that the word found is not the word sought.

5. A place-name with a feminine ending

The shape: Egyptian forms a city's goddess from the city's name plus -it. The result is a real goddess whose name tells you the town and nothing about her.

Worked example: anit. Budge states the rule outright: "the cities An, and Apt, and Behutet, etc., possessed goddesses of the city which were called Anit, and Apit, and Behutit."

This does not make them unreal — a city goddess is a goddess, worshipped, listed, drawn. But an entry built on the name alone is describing a place while sounding like it is describing a person. And the rule is productive, so an unknown number of census names are towns.

Why it is hard to catch: it is not an error at all, just a fact about the grammar that changes what an entry can honestly say.

6. A priestly office, read as a deity

The shape: temple personnel have titles. Titles get transcribed. A title held by a human being enters the list as a god.

Why it is hard to catch: the office is often named for a god, or contains one, and appears in religious texts in the same registers as divine names.

7. A named object, or a piece of architecture

The shape: Egyptian funerary religion names its furniture, and the names are grammatically identical to divine names.

Worked example: ur-heka. Budge's own index settles it on two consecutive lines — "Ur-heka instrument, 39" and "Ur-hekau, 130, 133, 595." The first is a ritual implement from the Opening of the Mouth; the second is the god, spelled with the plural ending. The census took the tool.

And the reason this form is worse than the others: in these texts, everything that has a name can speak. At the Door of the Hall of Two Truths, after the dead man has satisfied the forty-two judges, twelve parts of a doorway interrogate him in the first person — the bolt, the posts, the threshold, the socket, the lock, the floor, and his own two feet, each refusing passage until told its name. A bolt that says "I will not let thee enter in over me unless thou tell my name" is, on the page, behaving exactly like a god.

Why it is hard to catch: you have to read the passage. No property of the string distinguishes a door from a deity.

What actually catches them

Nothing clever. Read the passage the name occurs in.

Every one of the seven was found the same way: locate the name's position in the source, read a thousand words around it, and see what the sentence is doing. Hit counts do not do it — a hit count cannot tell six attestations from six copies of one fact. Variant generators do not do it; they are how form 4 happens. Adjacency does not do it either: two names printed sixty characters apart may be a roster, or may be alphabetical order in an index.

Two habits do most of the work:

Check the neighbours before trusting proximity. If the names around a hit run A, B, C in alphabetical sequence, the passage is a list and tells you nothing about who belongs with whom.

Do the deliberate check even when you expect nothing. After two objects turned up as false deities, this reference searched the door chapter specifically for more. It found none — only two of the twelve door-part names are transliterated Egyptian at all; the rest are given in English translation, so a scraper could only ever have taken those two, and one of them is neb-pehti. That search closed the question. A check that finds nothing still earns its keep, because it turns an open worry into a bounded one.

What this reference does about it

It does not delete them. A name that reached the census from somewhere deserves a page saying what it actually is, because the next person to meet it will need exactly that.

So a name identified as furniture, a tool, a title or a town is graded ⚠ audit with a verdict of identified — not a deity, and the entry sets out the evidence. Deleting would be tidier and less useful.

And the count reflects it. Names resolved this way clear the research queue exactly as written entries do, because what was needed was an answer, and the answer is that this is a door.