Nag Hammadi: The Monks Were Reading Them

Our file's timeline is unflattering to everyone: pages burned for fuel in 1945, texts circulating among antiquities dealers, Jean Doresse identifying them as Gnostic in 1948, most of the collection reaching the Coptic Museum in Cairo by 1956, a UNESCO project from 1966, and the complete Nag Hammadi Library in English only in 1977, thirty-two years after the find. It is very tempting to read that as more suppression, and our own file refuses to: the delay is more reasonably attributed to the sheer difficulty of conserving, photographing and translating fragile Coptic papyri, since work on ancient manuscripts routinely takes decades.

The core of it, as our file reports: the maker of the material world is not the true God. He is the Demiurge, called Yaldabaoth, or Saklas, which means fool, or Samael, which means blind god. He is born from Sophia's error, he is ignorant of the realm above him, and he announces that he is God and there is no other. He makes Archons, usually seven, one for each planetary sphere, described in the texts both as animal-headed beings and as psychological forces, whose job is keeping human souls in matter. And in several texts the serpent of Eden is a good figure: in On the Origin of the World it is sent by divine Wisdom, in the Hypostasis of the Archons the spiritual woman enters the serpent in order to teach Adam and Eve, and the Testimony of Truth asks what sort of God forbids knowledge. Our file marks all of this Tier 1 as text and Tier 3 as cosmology, which is exactly the right split.

It is worth resisting the phrase Gnostic library. Codex VI contains a Coptic translation of a passage from Plato's Republic alongside Hermetic texts. Codex XII contains the Sentences of Sextus, a set of ethical maxims that are not Gnostic in any sense. Our own file says this plainly in its Scholarly Cautions: not all the Nag Hammadi texts are Gnostic, the texts sometimes contradict each other, and reading them as a unified counter-theology imposes a coherence that may not exist. It goes further and repeats Michael Williams's 1996 argument that Gnosticism may be a modern scholarly invention rather than a thing that existed.

Hugo Lundhaug set out the shift in 2022. Most Nag Hammadi research, he writes, has been undertaken from the perspective of the texts' original authorship rather than of the time and place in which the surviving manuscripts were produced and used. The manuscripts are direct evidence of people who actually copied and read them, and for many of these texts they are the only such evidence. He adds a warning that is easy to miss and cuts deep: when a work survives in only a few copies, that hides the fluidity of its transmission and gives a misleading impression of textual stability.

In 2023 Paul Linjamaa published a study of the codices as physical artefacts, working from codicology, paratextual marks and scribal habits. His conclusion, in his own summary: the research supports the hypothesis that the Nag Hammadi texts had their origins in Pachomian monasticism, and shows how Pachomian monks used the texts for textual edification, spiritual development and pedagogical practices. It further finds that the texts were used for perfecting scribal and editorial practice, and that they were used as protective artefacts containing sacred symbols in the continuous monastic warfare against evil spirits. His own phrase for what this does is that it challenges preconceptions about ancient orthodoxy.

Our file notes that the Apocryphon of John appears three times across Codices II, III and IV, and calls the duplicates significant for textual comparison. It says the Berlin Codex's fourth copy confirms textual stability across multiple manuscript traditions. And in the same breath it records that the Codex III version is SHORTER and the Codex IV version LONGER. Copies of different lengths are evidence of fluidity at least as readily as of stability. Which of the two they show is the question, not the answer, and this is precisely the move Lundhaug is warning about.

And here is the sentence that should govern this whole subject, and it is our research file's own. On the claim that these texts are the truth, deliberately suppressed by a Church that knew it: this overstates the evidence. Suppression was real, systematic and documented, but the motivations were theological, political and institutional rather than necessarily conspiratorial. Then the line: reading them as the suppressed truth is as ideological as reading them as heresy. A document in a library like this one, refusing the most marketable version of its own subject, is worth more than any of the texts it summarises.

Our file draws a line from the Archon cosmology to modern prison-planet and simulation ideas, noting correctly that the Wachowskis have explicitly cited Gnosticism as an influence on The Matrix. There is now scholarship on exactly that move. Jan Kozák's 2023 analysis of gnostic films argues their cosmology bears a close similarity to the cosmology of conspiracy theories, and that the wave of such films around the turn of the millennium is part of the sociocultural development of the conspiracist mindset entering the mainstream. Which reframes the resemblance usefully: it is real, and it is evidence about the 1990s rather than about the fourth century.

Our file ends with a Corrections block from the campaign that repairs this library's bibliographies. Two ISBNs were changed, each with the same reason: the previous number failed its check digit. NEITHER OLD NUMBER FAILED ITS CHECK DIGIT. Both compute clean, and both resolve. The Pagels number pointed at Los Evangelios Gnósticos, the Spanish translation, so the real fault was a wrong-language edition. The Irenaeus number pointed at a 1977 Brill monograph ABOUT Irenaeus rather than Irenaeus's own text, so the real fault was a wrong work. Both replacements are defensible and one is clearly better. The repairs were right. The reason given for them was not.

That sentence appears thousands of times in this library, so it was worth measuring rather than generalising from two cases. Every correction line was harvested and every number computed, by a script left behind so the figure can be re-checked. The result is more interesting than the accusation. Of 2,307 correction rows, 1,679 of them, 72.8 per cent, genuinely did have a failing check digit: the stated reason is TRUE in the large majority of cases. In 545, about 23.6 per cent, the old number was perfectly valid and something else was wrong with it. In 83 more, 3.6 per cent, the old and new values are the same string, so a repair was logged and nothing changed. And every single replacement computes clean. THE REPAIRS ARE SOUND; THE REPORTING IS NOT. One fixed sentence is emitted for at least three different situations, and it is wrong about the reason in 628 of 2,307 rows, 27 per cent.