The phenakistoscope was the first known animation device to rely on rapidly displayed image sequences, and it dates back only to the first half of the nineteenth century. However, some much older pictures exist of successive moon phases arranged into circles, and these closely resemble phenakistoscope discs in format and logic. I recently had the idea of trying to set some of these illustrations in motion, and that’s what this blog post is about.
Back in the day, a phenakistoscope disc with narrow, evenly-spaced slits could be held facing a mirror and spun while the viewer looked through the slits at the disc’s reflection. The slits would expose the reflection for a split second each time the disc rotated a step forward in the image sequence, making the images appear to swap out for one another and producing an illusion of movement. Other arrangements were made as well—with separate discs for the slits and images, for instance, or with means of projection onto a screen, as with Eadweard Muybridge’s zoopraxiscope—but the principle remained the same: the images on the disc were somehow made visible only at set rotational intervals and concealed from view while the disc was moving between them. Today we can achieve the same effect digitally. Here’s an animation of a sequence from Muybridge’s Descriptive Zoopraxography (1893): plate 1, “Athlete, Horse-Back Somersault.”
There are fourteen images spaced evenly about the circle, so for each successive frame in my animated GIF I’ve rotated the image counterclockwise by one fourteenth of 360 degrees (which comes out to about 25.714 degrees). I’ve also shown some of the surrounding page in an effort to illustrate as clearly as possible what’s involved in contriving the illusion. You can see a zoomed-in animation of the same plate at Wikimedia Commons. For more early animations in this spirit, check out the Richard Balzer Collection—there’s some amazing and delightful stuff there.
In order for images to be suitable for phenakistoscopic display, they need to be evenly spaced around the perimeter of a circle, and they need to represent successive phases of motion at regular time intervals, with the action “looped” in a closed cycle with no beginning and no end. You might assume that nobody would have created a set of images fitting those parameters before the invention of the phenakistoscope itself. But you’d be wrong, and that’s where the moon phase charts enter the picture. The charts I have in mind represent changes in appearance over time using exactly the same format a phenakistoscope disc does. Of course, they weren’t intended to create an illusion of movement; they were intended strictly for reference and study as still images. But we can still animate them today as an exercise in what I like to call eduction against the grain.
Take the celestial map found in an Ottoman Turkish manuscript of the Zubdat al-Tawarikh (“Cream of Histories”) by Seyyid Loqman Ashuri, dedicated in 1583 to Sultan Murad III. The map shows twenty-eight images of successive moon phases spaced out in a circle running counterclockwise. Here’s an animation I’ve created from it in which each frame is rotated clockwise from the previous one by 12.857 degrees (one twenty-eighth of 360 degrees).
If we zoom in even closer and try to align the “moons” more closely with each other, we get a nicely-framed moving picture derived from a sixteenth-century representation of moon phases—somewhat wobbly and blurry, with artifacts of color halftone printing besides, but still pretty darn cool.
For comparison, here’s a GIF based on actual photographs of the moon passing through its phases, adapted from an animation at Wikimedia Commons.
A date of 1583 makes the Zubdat al-Tawarikh sequence four hundred and thirty-two years old, which is nothing to sneeze at. However, we can cast our net back further in time than that. A similar celestial map appears in the famous Catalan Atlas of 1375.
Here’s a close-up with the “moons” manually aligned—the damage at the center split is a bit distracting, but try to follow the gilded area as it passes from right to left and you should be able to experience a reasonable illusion of change over time.
The two moon phase sequences shown so far are impressive partly due to the large number of images involved—twenty-eight in both cases. Shorter sequences of eight successive moon images are easier to find, though, and they can yield striking effects as well. Here’s one such example: an astronomical calendar by John Somer, circa 1444 (Bodleian Library, MS. Ashmole 391(2), fol. 017r).
One especially promising astronomical chart from this standpoint is the “Cursus solis et lunae per signa singula” (“course of the sun and moon through single signs”) in Bayerische Staatsbibliothek (BSB) manuscript Clm 210, known as the “Salzburg Encyclopedia of 818.” It contains two different rows of sequential images. I think the outer one is supposed to show the movement of the sun according to some geocentric model, but the inner row definitely represents moon phases, and that’s what I’m going to focus on here. There are twenty-six separate moon images arranged counterclockwise, so I’ve created a twenty-six-frame animation with the source image rotated 13.846 degrees clockwise for each successive frame.And here’s our close-up—this is fun to watch, but it clearly isn’t working as well as the previous two examples.
The irregularities are due partly to uneven spacing of the moon images and partly to the fact that the illustrator didn’t try very hard to show the varying extent of lunar illumination; the moon is simply new, full, waxing gibbous, or waning crescent, with nothing in between, so it seems to “jerk” from state to state. However, the content of BSB manuscript Clm 210 is duplicated in ÖNB Cod. 387, an “Astronomisch-komputistische Sammelschrift” at the Austrian National Library. The corresponding illustration in that manuscript (scan #342) has twenty-eight moon images, so whoever produced Clm 210 might accidentally have left two of them out when making that copy. Here’s an animation from ÖNB Cod. 387.
When we zoom in, we still see the moons bobbing around like ping pong balls in a washing machine, but at least they hold steady enough for the eye to follow one of them through its phases, which this time are depicted as passing through progressive degrees of illumination.
Both specimens of the chart seem to contain different copying errors. BSB Clm 210 shows the zodiac rotated 180 degrees from where it should be, for instance, while ÖNB Cod. 387 had an extra circle painted in at the top of the sun sequence which was then colored over, leaving a conspicuous gap. With these divergent glitches in mind, I suspect both copies might have been based on some earlier, more carefully-drawn source—one in which the images were more evenly spaced and the lunar and solar cycles coincided more neatly. If that source could be tracked down, I assume it would yield even better phenakistoscopic results.
That’s as far back in time as I can confidently go with these moon-phase animations. I’ve animated the source images digitally here for online display, but there’s no reason a comparable effect couldn’t be achieved in a more traditional way with any of the examples we’ve considered so far—say, by printing a copy on paper, mounting it on a cardboard wheel, cutting evenly-spaced slits into it, and spinning it in front of a mirror.
But let’s push the envelope a little further. There’s an even earlier astronomical drawing in the same spirit credited to the Armenian scientist Anania Shirakatsi (AD 610–685), available online at Wikimedia Commons from a reproduction in the Soviet Armenian Encyclopedia. The caption identifies it as illustrating “phases of the moon,” but if that’s true, it does so in a way very unlike the previously seen examples. Robert H. Hewsen (“Science in Seventh-Century Armenia: Ananias of Sirak,” Isis 59:1 (Spring 1968), pp. 32-45, at 38) writes that Shirakatsi attributed moon phases “to the fact that the constant movements of the sun and moon cause them to change their positions in regard to one another, which thus results in the change of contacts between the light of the sun and the moon’s surface.” This illustration could have something to do with that, I suppose; perhaps it’s an attempt to show something similar to the two rings of images taken together in the “Cursus solis et lunae per signa singula.” But the caption could also be wrong, for all I know—maybe this is actually supposed to illustrate something entirely different, such as a lunar eclipse. In any case, there are eight apparently successive images, whatever they may happen to represent. They’re not very evenly spaced, but I’ve come up with a modified technique to accommodate them: for the animation shown below, I’ve rotated the image 53.5, 92, 140, 193, 240, 276.5, and 314 degrees counterclockwise as needed to hold the black circle consistently at the top of the image—that’s the point you should focus on as you look at it. I’ve assigned a consistent duration to the frames, but we could also adjust their timing to correspond to the varying rotational distances between them. If we wanted to convert this into a “real” phenakistoscope disc, we could cut slits into it spaced unevenly at each of the angles listed above to get a comparable effect.
—is a pretty good animation of Armenian astronomical drawings dating back at least 1,330 years. Maybe someone out there can tell me which one it is, and what exactly the drawings are intended to show.
Some other historical moon phase sequences don’t lend themselves to phenakistoscopic treatment even as well as the Shirakatsi illustration does, but we can still animate them in other ways. As two cases in point, consider the twenty-eight image sequence in Athanasius Kircher, Ars Magna Lucis et Umbrae, published in 1646, plate XXII (below left), and the thirty-six image sequence in the Selenographia of Johannes Hevelius, published in 1647, on page 183 (below right).
The Hevelius plate has its images arranged in a circle, but the orientation of the moon’s axis holds steady relative to the top and bottom of the page rather than to the center of the circle. To animate these images, I think we’re better off superimposing them “as is,” in the style of a straightforward flip-book, rather than rotating them as in a phenakistoscope.
The Kircher plate shows the moon’s axis changing orientation much as in our earlier examples, but the images are arranged into an ellipse rather than a circle, and the presence of a face complicates matters besides. There’s arguably no meaningful “center” around which we could rotate the ellipse, but here’s a flip-book style animation, like the one I made from the Hevelius plate.
I’ve been focusing so far on moon phases, but illustrations of other astronomical phenomena look like a promising source for primeval animations too, and I’d like to conclude with a non-lunar example. Here’s a famous plate from Christiaan Hugyens’s Systema Saturnium (1659), printed twice on pages 55 and 60, showing Saturn in its thirty-year orbit around the sun as viewed by a distant celestial observer (inner ring of images) and by an observer on Earth (outer ring of images).