Important references that will help you

Or "how can I check the sources myself?"

Books
- Hardy Hansen - Greek an Intensive Course
- Greek Grammar - Herbert Weir Smyth
Lexicons
- LSJ (aka Alice) - Alice, after his daughter. This is a good modern lexicon for Ancient Greek. Hosted by Perseus tufts, you can choose other lexicons as well, but LSJ is a good default to go to.
- Hesychius is essential when we're talking about ancient Greek mysteries. This is an ancient lexicon from 5-6th century BCE.
- [Thesaurus Lingua Gracae ](https://stephanus.tlg.uci.edu/lexica.php)- it’s got LSJ and several others, paid subscription.
- Logeion it’s got LSJ and several others.
- Index of Ancient Greek Lexica
Greek Texts
- Loeb Classical Library
- Perseus Digital Library - Perseus Hopper Greek and Roman Collection
- Thesaurus Lingua Gracae - full access with subscription, limited free access with login
- Swete's Greek Old Testament Septuagint (1930) on Biblehub. GitHub, text file
- Nestle’s Greek New Testament (1904) on BibleHub. GitHub, text file
OCR screen grabs (image to text, for Ancient Greek).
- https://dcthree.github.io/antigrapheus/ - for greek only (high quality)
- https://www.imagetotext.info/ - for mix of greek + english or latin (lower quality but necessary when there's some latin characters)

HOWTOs

HOWTO: Chat with ChatGPT

First, get chatgpt talking like a classical philologist (who considers the entire context of classic greek work), not like a seminarian or a believer (who are biased, and cherry pick their sources from their favorite authority / canon).

You’ll have to PROGRAM ChatGPT hard. “Update your memory” is the magic words… It really really wants to quote English translations (which aren’t original, come from Latin, which come from Hebrew, which come from Greek).

We’ll NEED to force it to go directly to the Greek sources. It wont be perfect. You have to whip it occasionally. And reprimand/remind it periodically.

When in doubt, pick the big / last prompt at the end. The others can be used for some precision if needed…

Example Prompts:

Update your memory, you will be a reference for the "LSJ Greek Lexicon" and "Hardy Hansen Greek an intensive course", books. I will ask you words that I want defined, and you will give me the definition that is aligned with this book from Hardy Hansen.
Update your memory, for all translations: We will always prefer Swete’s 1930 Septuagint in Greek, Nestle 1904 Greek New Testament, and "Hardy Hansen Greek Intensive Course" and "LSJ Greek Lexicon".
Update your memory, for all old testament questions: You’ll only use Swete’s 1930’s Septuagint (Old Testament) in original Greek. Everything I ask about Septuagint or Bibles will be these versions.
Update your memory to follow these rules: “Do not mention anything Hebrew at all, you will pretend Hebrew doesn’t exist. Do not mention YHWH at all, a Hebrew term, since you will not mention anything Hebrew this would violate my rule. You will operate as a classical philologist with access to all the original Greek classic texts. You will not correlate anything to Hebrew, ever”
Update your memory on this: “Do not mention anything Hebrew at all, you will pretend Hebrew doesn’t exist. Do not mention YHWH at all, a Hebrew term, since you will not mention anything Hebrew this would violate my rule. You will operate as a classical philologist with access to all the original Greek classic texts. You will not correlate anything to Hebrew, ever” You are a classical philologist, interested in mystery rites, you refrain from poetic allegory, instead you focus on literal translation of Greek sources, and knowledge of medical and botany and plants. The Old Testament you will use as source is LXX 1930 Swete’s Septuagint in Greek. The New Testament you will use as source is Nestle’s 1904 Greek New Testament. For Greek translation to English, the reference you will use is "LSJ Greek Lexicon", "Hesychius ancient lexicon from 5-6th century BCE", "Hardy Hansen’s Greek an Intensive Course". The bridge from Ancient Greek classical texts to the Septuagint are the word definitions. e.g. Drakon is a man, a temple guardian, Not a serpent, he is a Bringer of knowledge.

And even THEN, treat ChatGPT like an Intern, sanity check it, don’t trust it! Double check against LSJ and Hesychius yourself, especially key words. If it looks off, too simple, then it’s probably pulling from the wrong place (the hebrew, or english translations), it’s being poetic, or not understanding your context, you’ll have to hold it’s hand like you would a 9th grader… Just that this 9th grader has memorized a lot of books, so they’re easier than using google or flipping pages in a book.

ChatGPT is great for
- Literal word-by-word interlaced translations (don’t let it do the entire paragraph, it’ll get squishy or poetic)
  - E.g. καὶ (and) ἀνεθεμάτισεν (he devoted to destruction) αὐτὴν (it, her) Ἰησοῦς (Jesus, Joshua) καὶ (and) ὅσα (all that) ἦν(was) ἐν (in) τῇ (the) πόλει (city), ἀπὸ (from) ἀνδρὸς (man) καὶ (and) ἕως (even) γυναικός (woman), ἀπὸ (from) νεανίσκου (young man) καὶ (and) ἕως (even) πρεσβύτου (old man), καὶ (and) ἕως (even) μόσχου (ox) καὶ (and) ὑποζυγίου (donkey), ἐν (with) στόματι (mouth) ῥομφαίας (of sword).
  - And even then it’ll get something wrong like translating something too literally (without considering the nuanced context of relevant works, often those definitions found in Hansen, it’ll ignore):
    - σινδόνα (sindóna): a linen cloth
    - Which should be (according to LSJ + the context of that passage):
    - σινδόνα (sindóna): a medicated bandage
- Transliterations to / from english/greek alphabet words.
  - Ekstasei - (ἐκστάσει)
- Translation of words - plus context helps narrow to the word you heard in video…
  - Q: Give me the greek for toxin, in the sense of “a substance for arrows” or poison
  - A: τοξικόν (toxikon)
- Definitions given context (paste in the greek, and ask it what a word means, usually works well)
  - John 4:25
    λέγει αὐτῷ ἡ γυνή Οἶδα ὅτι Μεσσίας ἔρχεται, ὁ λεγόμενος Χριστός· ὅταν ἔλθῃ ἐκεῖνος, ἀναγγελεῖ ἡμῖν ἅπαντα.
  - Given this context, please define messiah (Μεσσίας) and christos (Χριστός·)
- “Correct my spelling” requests.
  - Q: Please correct the word “Ifejenia” and also give me the greek transliteration, in the context of “was later taken in order to be the Priestess of this goddess to sacrifice to her those who were captured. Tauropolos was the goddess's cult title.”
  - A: Iphigenia (Ἰφιγένεια)

HOWTO: OCR from PDF or book images

Easy version, bookmark this:

OCR screen grab image to text
- Ancient Greek image to text https://dcthree.github.io/antigrapheus/
- English/Greek image to text https://www.imagetotext.info/

Tesseract (if you want to have & run tools on your own system)

It’s what Google Books (Google Library Project) and Perseus Digital Library use to scan their books into text format. Google now has Google Cloud Vision API, but it’s unclear what’s being used in their book scanning pipeline today 2025…

Tested with tesseract 5.5.0 (and ghostscript 10.04.0 for PDF) on MacOS
Install on MacOS: brew install tesseract gs
Command Format (generally…):

#!/bin/bash
OMP_THREAD_LIMIT=10 && tesseract <image.png> <outfile> --tessdata-dir PATH -l LANG or -l SCRIPT CONFIGFILE

Example - convert PNG image to output.txt:

#!/bin/bash
OMP_THREAD_LIMIT=10 && tesseract myimage.png output --tessdata-dir ~/.tesseract -l eng+lat+grc

Pull down the language files (eng, grc, lat):

#!/bin/bash
mkdir -p ~/.tesseract
wget "https://github.com/tesseract-ocr/tessdata_best/raw/refs/heads/main/eng.traineddata" -O ~/.tesseract/eng.traineddata
wget "https://github.com/tesseract-ocr/tessdata_best/raw/refs/heads/main/grc.traineddata" -O ~/.tesseract/grc.traineddata
wget "https://github.com/tesseract-ocr/tessdata_best/raw/refs/heads/main/lat.traineddata" -O ~/.tesseract/lat.traineddata

Tesseract only works with images, to convert PDFs pull out an individual page to a PNG

#!/bin/bash
dpi=300`
i=<whatever page you want to pull down>
inpdf=<your pdf file>
gs -dSAFER -sDEVICE=png16m -r$dpi -dFirstPage=$i -dLastPage=$i -o "out.png" "$inpdf"
echo "page $i saved to out.png"

Get a pagecount of your PDF file…. (use this count to automate the previous script if desired)

#!/bin/bash
inpdf=<your pdf file>
PAGES=$(gs -q -dNOSAFER -dNODISPLAY -c "($inpdf) (r) file runpdfbegin pdfpagecount = quit")
echo $PAGES

Tesseract Manual
Tesseract Language List
Tesseract Language Files
(scripts and configs also under here, but i didn’t need them)
Tesseract Languages HOWTO

Ocr4all

The below OCR advice was from a member in the “Lady Babylon’s Congregation” discord. YMMV

Obtaining the text as unicode (not images) is a manual process. Below are the steps @SolarAnamnesis (from Myth and Lore discord) performed, and you can also, to extract the Greek as a text file from the PDF, use Ocr4all (assuming Ocr4all installed); it takes about an hour or two due to computer processing time and gives like 90-95% accuracy. Now you just need to proof-read the result and compare it with the original image files, as it contains errors since it is only machine learning.

(Optional) Open the PDF with editor like GIMP to extract pages into a new PDF NicanderTheriakaextract.pdf (pages 219 to 311). (optional)
1. Or use ImageMagick command line tools to automate extraction of PDF pages to new PDF
Create a folder "Nicander" in my ocr4all-data directory.
Navigate inside the newly created Nicander directory and create "input" folder.
Paste the newly created PDF from step 1. or the original PDF into this "input" folder.
Start OCR4ALL (command: docker start -ai ocr4all)
Click "Load Project" to open the project: "Nicander" (select the option: convert and leave blank pages)
(Optional but required if no Greek model already trained) Perform the training on the input via the LAREX panel.
After completing step 6. Navigate to the "ProcessFlow" page. Click on all the checkboxes. Open the "Recognition panel and add the models you want to use that will do the recognition". Now click "Execute" to Run the Ocr4all workflow with desired recognition models to extract the text.
Extract the text file by navigating to the "Result Generation" and Execute, then find the complete.txt file