How Lost Secrets of Greek Astronomy were Rediscovered

Peter J. Williams

Recently the research team I lead made a number of discoveries relating to Ancient Greek astronomy. In a manuscript which has been radiocarbon-dated to the 5th or 6th century AD, we discovered early portions of the poet Aratus, the earliest texts of the mathematician Eratosthenes, some astronomical diagrams and, most recently, parts of the legendary lost star catalogue of Hipparchus.

St Catherine’s Monastery in Sinai, Egypt, the community that has preserved so many rare Ancient Greek texts over the last 1,500 years.

Aratus (c.315–240 BC) wrote a Greek hexameter poem called the Phaenomena (“Appearances”), which was a popular text for teaching the constellations. We now have the earliest manuscript copy of some parts of that poem.

Eratosthenes (c.276–195 BC) was the first person to measure the circumference of the earth accurately. He also wrote the Catasterisms, containing the myths of how each of the major constellations came to be set as they are among the stars. We now have a copy which is over 700 years older than those previously known, which improves our knowledge of Greek myths.

Eratosthenes teaching in Alexandria, Bernardo Strozzi, 1635 (Montreal Museum of Fine Arts, Canada).

Hipparchus (c.190–120 BC) is widely regarded as the greatest of all ancient astronomers. Founder of the discipline of trigonometry, his discoveries include the measurement of the solar year to within an accuracy of six minutes. The rediscovery of some of Hipparchus’ lost star catalogue gives us access to the first ever measurements of star positions made by humans. It also has implications for the whole history of astronomy, since the levels of accuracy are surprising. Previously, the earliest star catalogue to have survived was a later work by Claudius Ptolemy (AD c.100–70), which turns out to have been less reliable.

Some of these discoveries were made by undergraduate students doing summer internships at Tyndale House, a Bible research centre in Cambridge, where I am Principal. Here’s how the discoveries were made.

The orrery, Joseph Wright of Derby, c.1766 (Derby Museum and Art Gallery, UK).

Over a century ago Cambridge had as residents a remarkable pair of twin sisters, Agnes Smith Lewis (1843–1926) and Margaret Dunlop Gibson (1843–1920). They were wealthy Scottish presbyterian widows, who learned many languages and travelled widely in the Middle East, often unaccompanied, making significant manuscript discoveries. Through travel to Egypt and subsequent negotiation with another scholar, Lewis acquired a manuscript which came to be known as the “Codex Climaci Rescriptus” (“The rewritten book of Climacus”). It almost certainly came originally from St Catherine’s Monastery in Sinai, Egypt, since some parts of the same manuscript were found there in 1975 under a floor which had collapsed during an earlier earthquake.

Agnes Smith Lewis and Margaret Dunlop Gibson, portraits by John Peddie, 1920 (Westminster College, Cambridge, UK).

The manuscript is called Rescriptus (“rewritten”) because it is a palimpsest – from Greek palin (“again”) and psaō (“rub away”) – so named because one layer of writing had been rubbed out and another written on top. Palimpsests were actually very common late in the first millennium of our era, since there was a shortage of writing material. Papyrus in Egypt had been overfarmed, and paper, invented in China, had not yet made sufficient progress to the West. Finally, it is called Climaci (“of Climacus”) after John Climacus, Abbot of St Catherine’s Monastery in the early 7th century, because the top writing in Syriac contained his literary work The Ladder of Divine Ascent.

A 12th-century icon depicting the 30-rung Ladder of Divine Ascent, as described by John Climacus six centuries earlier (Monastery of St Catherine, Mount Sinai. Egypt)

Right from the beginning it was known that there was writing underneath the Syriac. Much of this could be seen and even read with the naked eye. Other parts could be read with the assistance of an ultra-violet lamp. Even harder parts were made legible by applying a chemical reagent – a practice which has been abandoned for over a century now, since it damages manuscripts. With the aid of these methods Lewis and then others were able to decipher most of the pages. The underwriting turned out to be a large collection of about 100 pages of Aramaic from Palestine, written by Christians in the 5th and 6th centuries. These texts were biblical and theological. Then there were 27 pages in Greek from the same period containing biblical texts. But there remained about 9 pages which defied decipherment: scholars could see some traces of Greek writing, but they couldn’t make out what they said.

All this changed in 2010 when the Cambridge theological college to which Lewis had donated the manuscript sold it. The book was purchased by the Green family, owners of Hobby Lobby (the large chain of US craft stores), who donated it to the new Bible museum they were founding in Washington, DC. In 2012 I agreed that our team at Tyndale House would research the undertext of the manuscript on behalf of what would become the Museum of the Bible. The museum in turn commissioned various rounds of digital imaging, including multispectral imaging to help us read it.

Summer interns researching Codex Climaci Rescriptus at Tyndale House in Cambridge.

Between 2012 and 2017 we had 17 summer interns working together to read the manuscript using multispectral images. From the very first summer in 2012 we saw that the manuscript had more to offer. Jamie Klair, at that time a Cambridge undergraduate in Theology, noticed a line with a reference to the island of Naxos; he identified the particular text using the Thesaurus Linguae Graecae, the best database of all Greek literature. It appeared to come from a scholarly comment on Aratus’ Phaenomena. This text was later recognized to have come from Eratosthenes.

Jamie Klair and other interns presenting preliminary findings to scholars in 2016.

Over the years, with repeated rounds of imaging and more than 4,500 hours put in by interns, we made constant progress on the transcription of the text. Sometimes a student might spend a whole day just finding a letter or two. Gradually we built up a clearer picture of the manuscript as a collection of some of the most significant texts in Greek astronomy. But many pages were still too difficult to read.

This aquatic creature, perhaps the constellation Delphinus, emerged suddenly when Vasilis Kasotakis, an image processing scientist, struck on the right algorithm.

Then in 2017 the museum commissioned imaging and processing provided by a combination of the Early Manuscripts Electronic Library, the Lazarus Project of the University of Rochester, and the Rochester Institute of Technology. 42 images of each page were captured using different wavelengths and light filters and then programmers processed the images to make particular text more visible to researchers. In 2018 the image-processing specialists joined our textual scholars for a week at Tyndale House and we were able to give real-time feedback on the particular pixels or ink we wanted enhanced. Progress sped up.

One of the programmers, Vasilis Kasotakis, using Principal Components Analysis and Independent Components Analysis struck on exactly the right formula, so that suddenly a creature like a fish appeared on his screen where nothing significant had previously been visible. He let out an exclamation and everyone in the room crowded round to see his discovery. Here he is sitting in the middle smiling.

Vasilis Kasotakis smiling as scholars gather round to admire the emergence of an erased picture of an aquatic creature.

By the end of that week we knew that the manuscript consisted of four types of material:

    • Lines from Aratus’ poem the Phaenomena
    • Drawings of the constellations
    • Stories from Eratosthenes of how the constellations arose
    • Listings by Eratosthenes of the stars in each constellation

These were coordinated together, so that Aratus’ poem took the lead, and when it mentioned a particular constellation the relevant text by Eratosthenes about that constellation was inserted.

Then in early 2021 during one of the Covid lockdowns, I was preparing the manuscript for publication, and thought I should attempt one of the pages which we hadn’t yet deciphered. In the processed image I was using the upper Syriac writing was black and the Greek script I wanted to see was red.

The mu followed by nu and epsilon with a line over, meaning 55 degrees, which was the first clue that we had stumbled on Hipparchus’s lost star catalogue.

I was struck by the clear sequence μ νε, mu followed by a small space and then nu epsilon. Most of the letters in the Greek alphabet were too wide to fit in the space between the mu and the nu and there were very few combinations in Greek which would make any sense at all.

Then I noticed something which I’d seen elsewhere in the manuscript: a thin line above the νε. That’s the usual method in Greek manuscripts of indicating that the letters are to be understood as a number. But if νε was a number it would be the number 55, and that was far higher than the other numbers in the manuscript, which at the most gave the number of stars in a constellation. So what if it were an astronomical measurement? Could the mu be an abbreviation for “degree”? After all the Greek word for degree is μοῖρα (moira). Then I noticed a little symbol above the mu. This turned out to be the circlet commonly used in the Greek abbreviation for degrees: μ̊. As I looked at the page it was clear that there were more measurements. So far our team had been reading more-or-less known text from Aratus and Eratosthenes. Now it was clear we were reading previously unknown text and that we needed help.

I was referred to competent collaborators in Paris. Within a month they could give numerous reasons why the coordinates seemed to be those of Hipparchus himself and, not long after, could even show that the figures in the manuscript showed Hipparchus to be more accurate than had yet been appreciated. Three of us wrote these findings up in the Journal for the History of Astronomy and the discovery also featured in Nature and in many popular news outlets.

Planisphaerium Arateum sive Compages orbium mundanorum ex hypothesi Aratea in plano expressa (Aratus’ Star Chart, or the Assembly of the Cosmic Spheres Shown on a Plane according to Aratus’ Theory), Andreas Cellarius, 1661 (Biblioteca Nazionale Marciana, Venice, Italy).

Thus Codex Climaci Rescriptus turned out to contain five types of material:

    • Lines from Aratus’ Phaenomena
    • Drawings of the constellations
    • Stories from Eratosthenes of how the constellations arose
    • Listings by Eratosthenes of the stars in each constellation
    • Star positions from Hipparchus’ Star Catalogue.

It is thus an amazing compendium of astronomical and astrological material. But we’re only just beginning research on this compendium. There’s still more text to recover, which will require even more advanced technology, such as X-rays of the manuscript. There may also be more parts of the manuscript still in St Catherine’s Monastery in Sinai, where there are at least 160 palimpsests, most of which have not been studied. There’s probably significantly more Classical literature waiting to be discovered. There’s plenty of work for those who aspire to make a contribution to Classics.

The majority of the astronomical texts from Codex Climaci Rescriptus have been published by me and four co-authors in the Classical Quarterly. In our first footnote, we thank 52 individuals, as well as some organisations, for their help. Discoveries like this require the collaboration of large numbers of scholars. There’s still so much to do in Classical studies. ars longa, vita brevis, or in reverse, life is short, the craft is long.


Peter J. Williams is Principal of Tyndale House in Cambridge, the UK’s largest institute dedicated to the research of the Bible. Though his book Early Syriac Translation Technique and the Textual Criticism of the Greek Gospels (Gorgias Press, Piscataway NJ, 2004) never became a best-seller, his book Can We Trust the Gospels? (Crossway, Wheaton, IL, 2018) has been translated into eleven languages. He self-describes as a Biblical Philologist.


Further Reading:

Peter J. Williams et al., “Newly discovered illustrated texts of Aratus and Eratosthenes within Codex Climaci Rescriptus,” Classical Quarterly 72 (2002) 1–28 available here.

Victor Gysembergh et al., “New evidence for Hipparchus’ Star Catalogue revealed by multispectral imaging,” Journal for the History of Astronomy 53 (2022), available here.

Jo Marchant, “First known map of night sky found hidden in Medieval parchment,” Nature 18 Oxt. 2022, available here.