By Gordon Rugg
In this series of articles, we’re imagining that you’ve gone back in time, and that you want to produce the Voynich Manuscript as a hoax to make money. We’re looking at the problems and decisions you’d face, and at the implications of various possible solutions.
This article is about how issues involved in producing illustrations and a script for the hoax.
When you produce images and a script with the intention of hoaxing, you have three human weaknesses working strongly in your favour. Two of them are fairly well known, but the third isn’t, and it shows up over and again in research carried out by people who think that the Voynich Manuscript is actually a simple problem. Those things are:
- Confirmation bias
- The birthday problem
All of these problems are on show in a recent paper by Tucker & Talbert that featured in New Scientist and elsewhere this month. I’ll discuss it in brief within this article.
(Images courtesy of the Beinecke Library)
Patterns and pareidolia
Human beings are extremely good at spotting patterns. In fact, we’re so good at it that we can spot patterns even when they’re not really there. The classic example is the face of Jesus allegedly appearing in improbable places, such as tacos or slices of toast. We’re particularly good at spotting resemblances to human faces, and there are numerous rock formations around the world with names such as “The Head of the Emperor”. Here’s an example; one of the several Elephant Rocks.
This tendency to perceive significant patterns where in reality there are only chance resemblances is known as pareidolia. It’s a favourite element of conspiracy theories, such as this famous example, the so-called Face on Mars.
A second favourite element of conspiracy theories in particular, and of dubious reasoning in general, is the human tendency to favour evidence that’s consistent with our pet theory, and to ignore evidence that’s inconsistent with it. That tendency is known as confirmation bias. It crops up everywhere.
A significant proportion of articles about the Voynich Manuscript show this tendency. Typically, the author finds some pieces of evidence that are consistent with their theory that the manuscript was written in Bulgarian, or Basque, or Nahuatl, or whatever, and emphasises those consistencies, while saying nothing about the much larger body of evidence that’s completely inconsistent with their pet theory.
To anyone with some knowledge of linguistics, this is a reliable source of entertainment, though often combined with professional exasperation, as in this example:
In brief, the odd qualitative and quantitative textual features of the Voynich Manuscript are extremely different from anything that occurs in any known language, and I don’t know of any serious Voynich researcher who believes that the manuscript is likely to be written in an unidentified language as plaintext (i.e. without being encoded). A lot of would-be researchers, however, don’t appear to be very interested in finding out what’s already been discussed and discovered, so the list of languages claimed as candidates for the manuscript gets longer decade by decade.
The birthday problem
The biases above are a fertile source of dodgy research on their own, but they’re aided and supported by a less well known phenomenon. The birthday problem in its original form involves calculating how likely it is that two people in a room have the same birthday, given a particular number of people in the room.
On the surface, it looks like a harmless piece of statistical trivia. However, it has far-reaching implications, because it crops up in a wide range of non-trivial real world problems, such as calculating the likelihood of any two unrelated components happening to fail at the same time within an aircraft.
Human intuition is very bad at guesstimating the answer for this problem, and even people with a basic grasp of probability theory tend to get it wrong. If you ask someone how many people would need to be in a room before there was a better than 50% chance of two of those people having the same birthday, the answer you get tends to be a very large number. In fact, the correct answer is 23 people in the room. It’s much more likely than most people would expect.
Why does this have any bearing on the Voynich Manuscript? It’s highly relevant, because the Voynich Manuscript contains a large number of illustrations – several hundred – and a very large number of words.
By sheer chance, and by the birthday principle, it’s highly likely that if you select any language at random, you’ll be able to find quite a few matches between words in that language and words in the Voynich Manuscript. On the same principle, if you select a region of the world at random, there’s a high likelihood that you’ll be able to find quite a few matches between plants in that region and plants depicted in the Voynich Manuscript.
This is a classic amateur mistake with languages, and it’s something that has cropped up repeatedly with the illustrations in the Voynich Manuscript. It’s unfortunate that Tucker & Talbert appear unaware of this problem.
Anyway, that’s some background theory. In brief, if you produce plenty of illustrations and a script with plenty of odd features for your hoaxed manuscript, then there’s a good chance that the reader will fall prey to one or more of the biases above, and start selectively interpreting what they see as confirmation that there are meaningful patterns in there, even when they’re just seeing the equivalent of Jesus’ face in a taco. The next section goes into more detail about practical issues. They’re fairly simple, so the section is fairly short.
Choosing the illustrations
One very practical issue that a hoaxer would need to consider is how long it would take to produce the images. A decent quality page of an illuminated manuscript would take about a day to produce, depending on quality, intricacy, etc. A typical image from the Voynich Manuscript takes about an hour to produce. That’s a significant difference, so if you can get away with doing quick and dirty images like those in the Voynich Manuscript, then you can make a significant difference in your production times.
That doesn’t mean that hoaxes involving high-quality images aren’t economically viable – as I discussed in an earlier article in this series, the key issue is return on investment, and there are some cases where a commercially viable hoax involved very large investment in time and materials.
In the case of hoaxing a document like the Voynich Manuscript, however, there’s a type of document where low quality images would fit perfectly well, namely the notebook of a researcher such as an alchemist. That explains scruffy writing (the lines in the Voynich Manuscript aren’t neatly aligned, unlike most professionally produced handwritten documents of the time) and also explains the amateurish images. In addition, it would add to the potential market value of the book, since an alchemist’s notebook is more marketable than, say, a set of merchant’s accounts.
One thing that would drive the market value even higher would be mysterious images. The Voynich Manuscript has an interesting combination of very mundane images, such as a plant that’s almost certainly intended to be a water lily, and of very strange images.
If you have the biases listed above working in your favour, then you can probably get away with a very wide range of images; mundane images will be perceived as evidence that the document is real, and unusual images will be perceived as evidence that the document is about something unusual. You win either way.
Here are some images I created a while back, in my Ricardus Manuscript.
Image copyleft Hyde & Rugg
They probably look like real plants that occur somewhere, but I didn’t model them on any specific real plants; instead, I used a “pick and mix” approach based on a few leaf shapes, a few root shapes, a few flower shapes, and so on. The plants look plausible, so there’s a strong temptation to interpret the two other images as also being plant-related – maybe seed heads? Even when you know that they’re all fictitious, they still have a pareidoliac effect.
It’s a similar story with creating a script.
Choosing a script
As before, the safest strategy is a mixture of the plausible and the novel. If you create your hoax using a single real script, such as Byzantine uncial, then you’ll have a hard learning curve before you become fluent in it, and you’ll also have the problem that any mistakes you make will probably be perceived by experts as mistakes made by a forger.
If, on the other hand, you use a script on your own invention, containing enough real characters to look plausible, and enough invented ones to make it clearly unlike anything else in existence, then you’re on much safer ground. Because of pareidolia and confirmation bias and the birthday problem, there’s a pretty good chance that any oddities in the script will be explained away by true believers, and/or that someone will discover an obscure real script which contains a character looking like one of your invented characters. That’s what happened with the Kensington Runestone and the James Ossuary and numerous other cases.
At a logistical level, if you have any sense, you’ll make the script easy to write. Some scripts are easier to write than others. A classic example of a cool-looking script is Enochian, invented by Edward Kelley, the Elizabethan con man.
Here’s a sample of what Enochian looks like. (It sounds as good as it looks – there’s a grainy audio floating around the Internet of Enochian being read by Aleister Crowley, the legendary occultist described by the press as the Wickedest Man in the World; well worth listening to if you ever get the chance.)
Note all the serifs (twiddly, sticking-out bits). It’s a striking and distinctive script, but it’s also a pig to write if you’re in a hurry.
Compare that with Voynichese. Yes, some of the rarer letters are twiddly and awkward, but the most commonly used ones can be written swiftly and easily. I can write Voynichese faster than I can write English.
Image courtesy of the Beinecke Library
Here’s the script I used in my Ricardus Manuscript. All of the characters are easy to write, and most of them are similar to Voynichese.
Image copyleft Hyde & Rugg
So, inventing a new script, and using it consistently throughout a long document, isn’t particularly difficult.
Hoaxing illustrations and a script for a document like the Voynich Manuscript can be surprisingly easy, provided that you keep clear of any unambiguously anachronistic details, and that you use the right types of ink and paint. In the case of the Voynich Manuscript, the same ink type was used for hundreds of years across Europe, and the paint types were similarly widespread across time and space, so the job of a would-be hoaxer is comparatively easy.
Illustrations make good commercial senses to a hoaxer; they can hint at tantalising secrets hidden in the unreadable text, and can, in the immortal words of W.S. Gilbert, add verisimilitude to an otherwise bald and unconvincing narrative. Or, in the case of the Voynich Manuscript, add verisimilitude to something that might not even be a narrative…
The Tucker & Talbert paper is available online here:
More about pareidolia:
More about the birthday problem:
Confirmation bias, and human bias in general:
The classic place to start is in the work of Daniel Kahneman and colleagues:
Over recent years, though, there has been considerable debate about human cognitive biases. Probably the best-known researcher arguing for a different perspective is Gerd Gigerenzer:
The Ricardus Manuscript is a ciphertext that I produced; it’s still undeciphered.
The James Ossuary is a stone box bearing an inscription that translates into “James, son of Joseph, brother of Jesus”. It’s generally believed to be a modern hoax, but this is disputed by many.
The Kensington Runestone is a stone found in Minnesota, USA, that bears an inscription in runic characters. It’s generally believed to be a modern hoax, but this is disputed by many.
I’ve discussed these issues and others in my book Blind Spot:
I’m posting this series of articles as a way of bringing together the various pieces of information about the hoax hypothesis, which are currently scattered across several sites.
Quick reassurance for readers with ethical qualms, about whether this will be a tutorial for fraudsters: I’ll only be talking about ways to tackle authenticity tests that were available before 1912, when the Voynich Manuscript appeared. Modern tests are much more difficult to beat, and I won’t be saying anything about them.
All images above are copyleft Hyde & Rugg, unless otherwise stated. You’re welcome to use the copyleft images for any non-commercial purpose, including lectures, provided that you state that they’re copyleft Hyde & Rugg.