Given that it’s April Fool’s Day, I guess I should be cautious about unveiling anything readers could suspect of being too good to be true. Some of the animations I’ll be sharing below might seem to fall into that category; after all, we’ll be seeing some centuries-old works of art “come to life.” But my goal here is not to advance an implausibly astonishing claim of my own, but rather to debunk one which others have made, or at least implied. Some histories of animation state that “[t]he Voynich manuscript that dates back to between 1404 and 1438 contains several series of illustrations of the same subject-matter and even few circles that – when spun around the center – would create an illusion of motion.” In this blog post, I’ll show why this is true—to a point—but not very meaningful. The illusions in question are quite real, and they’re neat to watch, but they don’t shed any light on either the history of animation as a cultural practice or the abundant mysteries of the Voynich Manuscript itself. Indeed, comparable illusions can be made from any rotationally symmetrical source whatsoever, as we’ll see. The Voynich Manuscript is remarkable in many ways, but our ability to animate some of its illustrations isn’t really one of them, and it doesn’t belong in timelines of the early history of animation.
The general principle I want to discuss here can be expressed as follows: anything with n-fold radial symmetry can be rotated 360/n degrees per frame for n frames to produce a phenakistoscopic animation. So, for example, here’s an eleven-frame phenakistoscopic animation of a Greek dish at the Louvre depicting a siren, circa 570-560 BC (MNB 626, source image from Wikimedia Commons):
And an alternative animation from a more evenly-lit monochrome image of the same artifact:
In these animations, the bird looks vaguely as though it’s scratching the ground with its back leg, waggling its head about, and jutting its neck forward to peck at something. Of course, this effect doesn’t have anything to do with any effort by the painter of the dish to represent movement. It just results from minor irregularities in spacing combined with random variations in the painting of bird images that were supposed to be identical. I think we can still justify the animation intellectually as a test and display of rotational symmetry, something that plainly was important to the painter of the dish. Moreover, since we learn something about the symmetry of the dish by animating it in this way, I’d argue that what we’re doing also counts as a solid example of eduction. Perhaps that’s a moot point; after all, I think the aesthetic impact of the animation really justifies itself. And even the scratching, head-waggling, neck-stretching bird isn’t something contrived on a whim in the twenty-first century, since it actually proceeds from the symmetrical structure of the ancient artwork itself—an unintentional byproduct, if you will, exposed to view rather than created arbitrarily. At the same time, it doesn’t furnish any compelling evidence that an ancient Greek dish-painter had conceived of the principle of animating still images by exploiting persistence of vision. The fact that we can do this is a happy accident.
The siren dish isn’t even all that unusual in letting us create such an illusion from it. Here’s another phenakistoscopic display of symmetry, the subject this time being a cosmological image from seventeenth-century India.
And here’s a phenakistoscopic animation of a small phalera, first century BC, found at Manerbio sul Mella, now held at the Museo Civico dell’Età Romana, Brescia.
We get a similar illusion, once again—although this last example illustrates the problematic effect of photographic lighting on phenakistoscopic experiments with three-dimensional source material; ideally, the lighting would be kept the same while the artifact was rotated into each of its positions and rephotographed. The same observation applies to the next example, a “reel with representations of Pegasus and Chimaira, probably an ear-stud,” likely Etruscan, early fourth century BC, at the Louvre (Bj 1887, photograph by Jastrow, Wikimedia Commons).
Here I’ve created an animation with fourteen frames to match the 14-fold symmetry of the outer ring of images, but the inner ring displays 13-fold symmetry, so that ring appears to drift gradually clockwise over time (if I’d gone with thirteen frames, the outer ring would have appeared to drift counterclockwise instead). The figures don’t line up perfectly, but apart from that they’re quite consistent in shape; presumably they were stamped by embossing dies. The circular punch seen between the figures seems not to have been rotated relative to the center of the reel as the figures themselves were, so in the animation it appears to spin.
In each of the works we’ve examined so far, the artist’s original goal seems to have been to produce a set of identical images evenly spaced around a circle, and nothing more. The jitter in the animations, which gives a superficial impression of movement, results merely from minor inaccuracies in repetition. Sometimes, however, we run into early instances of radial symmetry involving figures that differ enough from one another to imply that the goal was not to repeat them identically each time, but to represent different phases of motion. Here’s one such example: a fourteenth-century Persian Kashan Minai pottery bowl that was recently offered for sale by a Frisian dealer and could have been yours for €1000.
When we display the four images phenakistoscopically, they yield a convincing animation of a person jumping up and down, or perhaps dancing. Apart from position, the figure is essentially identical in each image: same size, same hat, same clothes, same face.
A similar example comes to us from a Thracian bowl of the first half of the fifth century BC belonging to the Duvanlii treasure discovered in Bulgaria:
The four images arranged at ninety-degree intervals seem consistent enough to be depicting the same chariot. If the animated wheels don’t “turn” and the animated horses’ hooves don’t do quite what we expect, that could easily be chalked up to the artist’s limited understanding of movement. Otherwise, the horses do seem to be galloping in the animation, and the chariot does seem to whip frenetically from side to side as it dashes along a bumpy path at breakneck speed.
Could this effect have been intended from the beginning? Well, maybe. It would theoretically have been possible to achieve the same kind of illusion centuries ago by placing the bowl on a potter’s wheel, mounting a plate some distance above it with four narrow radial slits cut into it at ninety-degree intervals, setting the wheel in motion, and looking down through the slits, just as in a later phenakistoscope. But I’d argue against drawing any conclusions about the history of the concept or practice of animation based purely on the visual effects seen above, when we have no evidence that anybody in antiquity ever actually tried to do what I’ve described or even had the intellectual framework needed to think of it. Here’s a comparable animation I created from an Iranian bowl of the late twelfth or early thirteenth century at the Metropolitan Museum of Art (ten frames):
This could certainly pass as an animation of a Persian horseman at full gallop. In this case, though, it’s pretty clear that the intention of the artist was to depict a number of different horsemen, and not a single horseman in different phases of motion. The horses are painted different colors; the horsemen are shown wearing different clothing. On the other hand, the images do still depict the different horsemen in different positions, if only for variety and in no particular order. When we animate them, the individual differences in color and clothing aren’t as conspicuous as the continuities in overall form, and the randomly-sequenced changes in position give a vague impression of movements too rapid for the eye to follow. My “galloping Persian horseman” shows how convincing an animation we can produce from a set of images that were originally intended to represent not a single subject in successive phases of motion, but a consistent type of subject in varied phases of motion. The little jumping person and Thracian chariot shown above might both fall into the same category. I doubt anyone could prove definitively that these weren’t used for animation, or at least intended to portray a motion sequence, and it’s admittedly fun to speculate along those lines. But the animated effects we see could just as well be accounted for in terms of an attempt to depict individual figures in a variety of positions within a radially symmetrical structure. Even though the images look similar enough to represent the same specific figure repeatedly, there’s no reason to assume they actually do.
And that brings me back around to the Voynich Manuscript. In case you’re not familiar with it, this is a mysterious codex written in an unidentified and undeciphered script and illustrated with enigmatic pictures, including a number of circular charts, all inscribed on vellum that has been carbon-dated to the early fifteenth century. On September 4, 2013, the Wikipedia “History of Animation” page was updated to state (as I quoted earlier): “The Voynich manuscript that dates back to between 1404 and 1438 contains several series of illustrations of the same subject-matter and even few circles that – when spun around the center – would create an illusion of motion.” This claim was removed on June 1, 2014 as “possibly fraudulent,” but over the preceding eight months it had been copied word for word into many casual online histories of animation, so it’s still widely represented on the Internet as fact. The source citation for the statement on Wikipedia was a now-defunct YouTube link to the Naked Science episode “The Book That Can’t Be Read,” which as of March 2015 is viewable here and here, and also here as ORF Universarium documentary Voynich Code: The World’s Most Mysterious Manuscript. This piece, which I’ll refer to here as the “ORF documentary,” contains several phenakistoscopic animations made from pictures in the Voynich Manuscript. You might want to take a moment to go watch the relevant segment before continuing—it starts at the 8:25 mark here or here, or the 8:40 mark here.
There doesn’t seem to have been much said about these animations among serious Voynich investigators (of whom there are many), and the only discussions of them I’ve seen appear in the comments sections of blog posts here and here. The former discussion likens the images to mandalas and contains the remark: “either we assume that the VM is a late-19th century hoax, or the middle ages already knew about zoetropes.” The latter discussion combines similar thoughts with speculation about how the image sequences might practically have been animated in the past while remaining bound physically into the book. Gert Brantner also remarks there of the ORF animations themselves: “their examples are short & few in number.. either too much work to examine all the drawings OR the theory falls apart with drawings not shown. Who knows (or has the time to check).” In fact, there’s not much “theory” to fall apart in the first place, since the ORF narrator states only: “Some of the book’s pages also contain optical phenomena. If set into a spinning motion, these illustrations come to life.” But let’s take a closer look at what this actually means.
The first two examples presented in the ORF documentary, encompassing larger circles, are based on the leftmost portions of foldout leaves 67r and 69v (you can find facsimiles of all pages here). Here’s my own animation of 67r:
I’ve rotated the images 360/n degrees per frame here to match an n-fold symmetry, just as I did in the examples shown earlier in this post. In the ORF documentary, a slightly lesser number of degrees was used instead to make the peripheries appear to drift in a direction opposite that of the centers, but that was done arbitrarily and doesn’t reveal anything about the sources themselves. These animations show nothing more than that the images in question are nearly symmetrical. They may “contain optical phenomena,” but only in the same sense as the Louvre siren plate does; there’s no reason to suppose the circular charts were intended for anything but static viewing.
The remaining three examples in the ORF documentary are close-up animations of female figures arranged into circles marked with signs of the zodiac. Each zodiac sign is associated with around thirty figures, sometimes spread out over two charts on two pages, with two or three concentric rings on each; thus, the individual figures seem likely to correspond to days in the calendar, although nothing is certain when it comes to the meaning of the Voynich Manuscript or any of its parts. The ORF documentary animations are based respectively on the outer ring of the chart on 72v;
and the inner ring of 72v, with irregularly spaced images—I haven’t bothered to recreate that one myself because it doesn’t come close to producing any illusion of movement. These other examples are at best equivalent to my “galloping Persian horseman.” In 72v, some of the figures have different headgear, including a crown in one case; they aren’t plausibly interpretable as multiple images of a single person in different phases of motion.
In 70r, one single image shows a figure emerging on the right of the tube-like object, which isn’t consistent with a set of sequential images showing the phases of someone emerging from the left of it.
This isn’t a very steady animation, to be sure, but it sort of works. One single image shows the figure wearing some kind of special headgear, but this seems to have been added (later?) in darker ink; the images as drawn (initially?) in lighter ink all appear consistently interpretable as iterations of the “same” figure, while the darker-ink additions might be interpreted as an effort to disguise that fact. (The same may also be true of 72v, but the headgear on 73r is less distracting than the big crown on 72v.) The position of the arm varies, but seemingly not at random: when it’s thrust out behind the figure, it remains that way for multiple iterations, much as we’d expect in an animation.
So does any of this count as evidence that the images were intended for animation?
Not really. I’d say my “galloping Persian horseman” presents an even more convincing illusion of movement than this does, but in that case the different clothing and differently colored horses show pretty conclusively that animation wasn’t the goal and suggest that such effects can arise simply as a matter of chance. Moreover, the Voynich Manuscript contains many other sets of similar images of women arranged into circles, and most of those wouldn’t work as animations even as well as the examples we’ve looked at so far: the images are too different from one another or too unevenly spaced. All the illustrations of this type were presumably intended to serve similar purposes, so if most of them don’t work as animations, it’s unlikely that this was the original purpose of any of them.
On the bright side, there’s no shortage of historical images out there which we can use to create phenakistoscopic animations just as visually impressive as the ones derived from the Voynich Manuscript—it’s just that they don’t tell us much about the past. I’ll leave you with one final animation of a miniature from Vaticanus gr. 1291, dating from between AD 813 and 820 and cited as a possible source of inspiration for the Voynich zodiac charts.