On the coast of Greece south of Athens are many places to see a beautiful sunset, shining across the sea. At the end of the peninsula, you also can see an unusual combination: As the full moon rises over the Aegean, the sun sets simultaneously in the gulf leading to Piraeus. The silvery shining waters lead from the eastern horizon toward you, and the rosy shining waters connect you to the western horizon. It feels a very privileged spot, as if something special is going on — and you are in the center of it.

     I think that this rising and setting might have impressed the ancient Greeks, too, because the view is best at the Temple of Poseidon, built about the same time as the Parthenon in the middle of the fifth century B.C., and perched on a rocky “land’s end” promontory jutting out into the sea like a giant breakwater. As the temple’s elongated shadows reach out toward the moon, the red-tinted Doric columns create a frame, making the moon appear enormous. Soon the sun has set and the moon has risen higher in the sky, but for a few minutes the convergence of shining streaks produces a spectacle — perhaps not on the order of an eclipse, but ranking up there with, say, the view from Delphi.

     The thunderstorms of the afternoon had disappeared. I got soaked twice during the day along the North Rim. In most places in the American Southwest, you can see little rain squalls coming, just by looking around the horizon in the direction from which the wind is blowing. But on the canyon rim, at nearly 2,400 meters (7,900 feet) elevation, you have to look down instead: squalls rise up out of the depths of the Grand Canyon, something like an eruption from a boiling cauldron that the witches forgot to stir.

     All of that well-heated air from summer sunshine in the canyon rises as updrafts — and runs into the moist air flowing north from the Gulf of Mexico during the monsoon season. A common result is a little turbulence that degenerates into patchy lightning and thunder. With the cooling of evening, the clouds often disappear, making possible a clear view of the western horizon.

     Walhalla Plateau is a peninsula-in-the-sky at the southeastern corner of the canyon’s North Rim. Cape Royal juts, promontory-like, south into the Canyon from the plateau, just as the temple’s site juts into the Aegean from the tip of the Greek peninsula. You have to imagine the Grand Canyon full of water to complete the analogy. The southern tip of Cape Royal is open and flat, except for the sagebrush that dominates the landscape. Scattered piñon pine and juniper trees border the mesa top; more rain falls where the storm clouds crest the rim, so that the sage represents a rain shadow of sorts.

     Seen from Cape Royal, sunset occurs over the Hualapai Indian Reservation on Coconino Plateau to the west, almost fifty miles away. And the moonrise is often over the Hopi Indian Reservation, on another high plateau about the same distance to the east. Thanks to the distances involved, the horizon is almost as flat as that seen from the Temple of Poseidon — but no shining seas are currently available to accentuate the contrasts, to connect you directly with the sunset or moonrise. Those yawning gaps are instead filled with warm colors and elaborate rock formations that cast long shadows.

     Angel’s Window is among the many thin ridgelines of rock that extend out, like free-standing walls, into the Canyon for a short distance from Cape Royal, before abruptly terminating in a cliff. They are miniature peninsulas on the fringe of the big one, shaped like vertical fins and possessing some sage and pine atop their cracks and crevasses. A large chunk of limestone fell out of one of those thin walls but, surprisingly, without collapsing the top of the ridge. And so a “window” in the rock was created. I remember looking up from the Colorado River, on one of my float trips through the bottom of the Grand Canyon, and seeing the large triangular opening in the cliff face, blue sky showing through it. Up at Cape Royal, tourists walk atop the window and look down. Or stand back on the plateau to the west and see the Colorado River framed by the window.

     Might this window function as a spotlight, like the one west of Delicate Arch? No, I reluctantly concluded. Only as it is setting can the sun be seen through the window — but it requires that the viewer be in the bottom of the Grand Canyon, standing in the right place. And the viewer would not be able to see any sign of a “spotlight” on the ground nearby. The “spot” created by the setting sun shining through the window is too diffuse to be detected, because the many miles between the window and the relevant areas of canyon bottom serve to smear out the shadow edges beyond recognition.

     And it was such a good idea; I could have used a second example of the Delicate Arch drama.

     Spotlights like the one at Delicate Arch are, of course, merely circular shadows. The direction of a shadow is a matter of the time of day and the time of year; the combination of sunset and summer solstice made that opening opposite Delicate Arch frame a shadow on the pedestal of Delicate Arch. While solstice shadows and sightlines are the usual preoccupation of archaeoastronomers, what about shadows and sights when an eclipse is due? Eclipses do not favor certain points of the compass (they can occur with the sunset anywhere between southwest and northwest), so no sightline is likely to be special to eclipses. But might sunset shadows point toward moonrise on special occasions?

     Hmm. Does your own shadow reach out and “touch” the rising moon just prior to an eclipse? When that idea first occurred to me, I remember speaking firmly to myself, something to the effect that I was getting a little too theatrical in my scientific musings, that my brief career backstage running the lights for a high school ballet production must have gone to my head. But I didn’t forget the notion of “touching the moon” with a shadow, connecting in a manner like those silvery streaks facilitated by rippling waters.

     Which night might prehistoric people have celebrated as the “full moon,” if fullness is so equivocal? My guess is that “full moon” (besides being called something else) was originally defined as the one evening of the month when the moon rises just before sunset — and so sits on the eastern horizon looking especially dramatic, colored by the sunset, with elongated shadows directing your attention toward it, and looming large (for some reason hidden inside our brain circuitry, the moon or sun sitting on the horizon looks a lot larger than when high in the sky).

     The previous night, the moon rises an hour before sunset and seldom gains the coloring; it stands high in the sky, white and small at the time of sunset. The night after the full moon, the sunset colors are often gone by the time that the moon (still looking quite full) appears on the eastern horizon. So on one night of this sequence, the moonrise is much more noticeable than usual. Before I could correct myself for relying on drama again, I remembered (modern geometry speaking again) that moonrise-just-before-sunset was an essential setup for a lunar eclipse later that evening. That definition of “fullness” would, it so happens, be one that facilitated eclipse forecasting.

     But, since lunar eclipses don’t occur every month, which month’s full moon portends an eclipse? How do we sort this out? Or rather, how might a prehistoric people have sorted it out?

     You need a good unobstructed path to play this game. The shadow becomes dimmer at the same time as it grows longer, and it slowly changes direction, sweeping slightly to the south. That’s because the sun is angling toward the northwest as it sets, sweeping those long shadows toward the southeast.

     What’s the relationship, if any, between shadow directions and lunar eclipses? The North Rim has no handy library with canons of eclipses and angles of moonrise and sunset to consult. But the little computer in my daypack had the programs for calculating the angles. Though I would like an excuse to visit the Grand Canyon at each full moon for the next year, I really could run time forward and backward in the computer, simulating what I’d see each month.

     Hmm. Before time measurement, they probably used the height of the moon at sunset as a simple way of ruling out eclipses. If the moon is more than a few diameters off the eastern horizon when the sun is sitting on the western horizon, a lunar eclipse is unlikely.

     It’s Method #4 (“High Moon is Safe”). If you have a prediction scheme such as counting by sixes, this moon height observation at sunset will provide a way of ruling out an otherwise predicted lunar eclipse, rather like you might look out the window in the morning to check the validity of yesterday’s weather forecast. You’ll be able to rest assured that the prayers will be answered, and perhaps gain some credibility at the expense of your competitors in the prophet business, who are raising the usual alarms on the sixth month following an eclipse. You can just look calm and assured, going around telling people that you know an eclipse won’t happen, whatever those other prophets claim, because you have a better pipeline to the Almighty.

     The method seems simple enough to have been discovered in prehistoric times. I certainly wasn’t the first to discover it in modern times as it is mentioned in passing in the very book that, a quarter-century earlier, originally stimulated my interest in eclipse forecasting:

     For about four centuries, we have known that the earth casts a cone-shaped shadow, the umbra, in the same direction as the observer’s shadow. This shadow cone rises unseen in the east, just as the sun sets in the west. Indeed, the observer’s sunset shadow is sometimes a tiny part of the cone, a little dim bump on the right side of it.

     An eclipse occurs when the moon’s eastward orbital motion during the night carries it left, into that conical shadow (well, at least as seen by observers in the Northern Hemisphere). But the prescientific observer need not know that in order to associate “touching the moon with your shadow” with an eclipse. So sunset shadows pointing near the rising moon (within a few diameters to its left) constitute Method #5 (“Touching the Moonrise”) for eclipse forecasting, once again “entry-level” in its simplicity for the prospective prophet.

     The method doesn’t always work (the moon’s path sometimes goes just above or below the umbra) but every eclipse will have been preceded by such a coincidence: the moon has to be close to the cone-shaped shadow at sunset to stand any chance of moving into the shadow within the next several hours. The scheme, particularly effective for those lunar eclipses that happen in the several hours after sunset, serves to warn of those lunar eclipses that happen when many people are still awake, rather like the pinhole methods give an hour’s warning of a total solar eclipse.

     Wider separations between moonrise and sunset shadows may also be associated with eclipses later in the night, but the judgment becomes complicated, making it hard to improve on the probability suggested by the counting-by-sixes prediction. With Method #1 (“Clenched Fist Counting”), you may know that there is a substantial probability of an eclipse; with Method #4 (“High Moon is Safe”) or #5 (“Touching the Moonrise”), you may be able to rule it out (as in the case of sunset before moonrise) or to raise the alarm (when the last shadows point at a just-rising moon, as must have been the case in Jamaica in 1504).

     Now that’s what they might have been doing at the Temple of Poseidon — using the sunset shadows of those Doric columns, stretching out across the marble floor and meeting the shining streak of sea, leading from the moonrise.

     The observer stands at the base of a western column, sights along the column’s shadow leading across the floor to the eastern columns, and judges if it lines up with the streak of silvery sea arriving from the rising moon — or if it produces too much of a dog-leg-like angle. Checking that sort of alignment, seeing if two straight-line segments convincingly merge into one straight line, is even easier than tracing one’s shadow outward toward the moon. Evolving a criterion for how-close-is-close can be aided by the columns themselves, such as requiring that the moonrise be within a column’s width of the sunset shadow.

     So is the Temple of Poseidon an architectural solution to eclipse forecasting: tall columns, built in a good locale (long views both east and west)? Suppose that the tall stones of Stonehenge were for the same purpose, the priests sighting down a stone’s shadow to see if the moon rose at its very end?

     You can often solve shadow alignment problems by looking directly at the setting sun, getting your line of sight from the sun itself rather than a shadow. Trouble is, I mused, how can you do that if you also have to look at the rising moon? Yes, I know that you can turn around, but how do you know that you’ve turned exactly half a circle, short of using a modern surveying instrument?

The low-tech solution (indeed, no-tech — just naked eye and found objects) is to use two observers, standing some distance apart from one another. The Eastern Observer stands still while the Western Observer walks around until the rising moon is located just behind the Eastern — and then remains rooted to the spot, while the Eastern Observer (continuing to stand still) sights past the Western toward the setting sun. If the setting sun is indeed behind the Western Observer, then the observers must be on the line from sun to moon, with no dogleg angle.

     The moon “touching” the Eastern Observer’s outstretched arm could also qualify — and thus provide a measure of the offset between moonrise and sunset shadow line (given some standard number of paces between the two observers). Like the within-a-column’s-width criterion for matching up a Doric column’s shadow with the moonrise, this would allow a tradition to develop concerning when to raise the alarm and when to let an ambiguous situation pass. It would give a clear yes-or-no answer (which always helps to reduce arguments!).

Now it seems unlikely that prescientific peoples would have formulated the rule as a “straight line” relationship. They’d have personified things, if folk culture is any clue. They might have called Eastern Observer the “Sun Priest” because he watched the sunset, called the Western Observer the “Moon Priest” (or some such). They would have watched for those occasions when the sun “touched” the Moon Priest in the same manner as the rising moon had “touched” the Sun Priest. Symmetry, no less.

     The Anasazi have living descendants who think that eclipses are bad luck, cause the deaths of children, and cause crops to fail. I wouldn’t be surprised if the Anasazi were similarly apprehensive about eclipses. Mentions of eclipses in the Pueblo literature are rare, but fear might have made eclipses taboo subjects for conversation, unlikely to be mentioned to an anthropologist or missionary. But there is a revealing story about a practice of two priests, reported by the anthropologist Florence H. Ellis from a pueblo in New Mexico, that strongly suggests (at least, to me) a procedure for eclipse prediction:

Each evening, they note the place of sunset and moonrise: “The moon is believed to travel between the north and the south just as the sun does but at opposite seasons so that their paths cross at one point. The matter of prime importance to these calendar priests... is to observe Sun and full Moon exactly when they most closely approach each other [emphasis added], a problem duplicating that of the Zuni....”

     With a somewhat different metaphorical translation allowing some ambiguity between the setting sun and a “sun priest” (perhaps the same word was used for both?), this might correspond to my Two Priests method: the full Moon approaching the Sun [Priest] and the Sun approaching the Moon [Priest]. Ambiguity has its uses, even for eclipse prediction.

     A certain duality between sun and moon, between day and night, between winter and summer is deeply embedded in Pueblo concepts. They conceive of an underworld that is a half-year in advance of the real world — except that, in the underworld, the moon may take the sun’s role. You can see where they might get the idea of the full moon playing the sun’s role in the underworld: in the winter, when the sun makes a low arc across the southern sky during the day, the full moon is making a high arc that same night — indeed, tracing a path similar to that taken by the sun in the summertime. And in the summer when the sun is high in the sky, the full moon traces a low arc like that of the winter sun. It would seem that an underworld notion might be helpful in evolving away from flat-earth models, toward schemes that, whatever their deficiencies, were handy for eclipse forecasting.

     The Two Priests method is likely to be a refinement of a one-observer method — at least, I can imagine stumbling into the shadow-touching-the-moon scheme while I cannot imagine any two people discovering the choreography associated with this method, absent some predecessor methods. Still, either is a relatively simple method that makes no demands on geometrical understanding — it just requires that you have a reasonably flat horizon to the east and west, and develop some rituals for watching full moons at sunset. Particularly if used in combination with the clenched-fist forecast or a magic number scheme, the sun and moon lining up might have provided a reasonably accurate warning of an impending lunar eclipse.

     Down along Unkar Creek are the thousand-year-old ruins of the winter homes of the Anasazi, probably the same families as lived up here on the Walhalla Plateau during the summer. They went south for the winter, down to where it is warmer. When the North Rim is buried in massive snowdrifts, the bottom of the Grand Canyon is merely a little cool and rainy.

     Such is the difference that elevation makes. The bottom of the Grand Canyon was a good place to shelter from the winter weather, with nuts to gather and game to hunt. At Unkar Delta near the river, the major farming area for the Unkar Creek people, archaeologists find lots of bones from bighorn sheep, deer, and rabbits. Even in the days before rainfall improved enough to allow farming, the local tribes probably did a lot of wintertime hunting and gathering in the bottom of the canyon.

     Downstream of the confluence with the Little Colorado River at Mile 61 below Lee’s Ferry (the usual place where river trips begin), the Colorado River’s narrow gorge opens out for about ten miles, takes a U-turn around Unkar Delta at Mile 73, whereupon the canyon walls close back in. The Anasazi farmed all along the stretch of more open river bottom, irrigating their maize, squash, and beans with water hauled from the river.

     On my third trip down the river, I saw a particularly long and spectacular lunar eclipse just after visiting Unkar Delta, in fact. That night, I was across the river, atop a minor hilltop with a good view of the expanse of Grand Canyon. The hilltop was distinguished by a minor ruin. Though it was only four walls made of piles of flat stones, the Cardenas Hilltop Ruin was something of an enigma. It seemed too far away from the agricultural sites (and drinking water) to be a habitation. Though a nice site for a guardpost because of its long views, it was really too large. Given the eclipse in progress, I couldn’t help contrasting it to Stonehenge, wondering if it was built there with some astronomical purpose in mind.

     Tracking the sunrise’s movements along the horizon, from its southeast extreme of winter to its northeast extreme of summer, was the first possibly prehistoric scheme that occurred to me while waiting for totality to end. Compared to the featureless horizon of Salisbury Plain, the bottom-up view from the depths of the Grand Canyon is far better for tracking the seasons. The hilltop ruin would have been just the place for the Unkar Delta people to “keep a calendar” for planting time. Each morning in winter and spring, the sun rises further north — until, at the summer solstice, the sunrise turns around and heads south again. One of those notches in the South Rim could be the Unkar Delta peoples’ reminder: When the sun finally rises on that notch, it’s time to plant corn.

If you can’t have a nice level sea horizon, better to have a jagged one, I thought. Mountain peaks and the notches between them function nicely as markers — so long as you maintain a customary point of observation, such as your front doorway. Or a favorite hilltop.

     I remember examining the eastern skyline in the moonlight. The Palisades of the Desert is an expanse of canyon rim with a vertically scalloped cliff. A regular series of promontories stick out, like folds of a hanging curtain — or, I suppose, the Doric columns on a Greek temple. All through the spring and summer, the sunrise would move along those notches. The Anasazi could have easily given the notches names, just as we name the months.

     And so I counted the number of notches in the skyline as best I could in the fading moonlight. Between the approximately due-east sunrise of the spring equinox and the northeasterly sunrise at the summer solstice, the sunrise position would shift sideways past at least a dozen easily identifiable markers. A dozen in three months means about one every week, on the average. That’s a pretty detailed calendar for agriculture, as you don’t need to count down the days.

When I scanned the southeast horizon where the sunrise should be in autumn and winter, I found that I couldn’t see the South Rim because a large butte inside the canyon obscured my view from the hilltop ruin. The butte wasn’t more than a few kilometers away and rose up rather high in the sky. That means that sunrise in the winter must have been pretty late. Might there be a special notch for the winter solstice sunrise, I mused, maybe in that nice V-shaped notch?

The winter solstice is the focal point of the Hopi’s high religious festivals, I remembered. For a week or so, the position of sunrise changes so imperceptibly that people speak of the sun “standing still.” Then it begins rising somewhat farther north each day. From Cardenas Hilltop Ruin, I estimated, it would reach the Palisades by spring and then continue through that series of notches during the springtime. The Hopi are known for spreading their bets by planting at different times in different places, so perhaps each one of those Palisade notches marked a time to plant. I wondered briefly about a big promontory with a pinnacle projecting from it, Comanche Point — what date might it correspond to?

     From where the Hopi tribe now lives just to the east of the Grand Canyon, they use the position of sunset over mountain peaks to mark the winter solstice, the time for religious festivals and planting corn, etc. The Indians who lived in Puget Sound could have easily done the same thing, what with the sunrise skipping from one mountain top to another in the Cascades as the weeks go by — and the sunset traversing the peaks of the Olympics as well (it’s between the twin peaks of The Brothers at the equinoxes, at least from Seattle). If you’ve named each peak and notch, as seen from some customary viewpoint, you can often infer “today’s date” to within a day or two of what a modern calendar would tell you (except near the solstices).

     After my river trip, I constructed a horizon calendar for the view from the hilltop ruin — not by spending a year there watching the sunrise, but with the aid of photographs, a topographic map, and the computer. Comanche Point turns out to be at the end of April.

     The diligent observer experiences great difficulty at the extreme positions because the position of sunrise changes so slowly from day to day. If you want to celebrate the turnaround, and your method can’t spot the reversal until a few days after it actually starts, then your winter solstice celebrations will occur — Christmas Day!

     That leads me to suppose that the megalithic monuments emphasizing solstice sightlines were for some purpose other than an agricultural calendar — a purpose that, incidentally, needed such solstice sightlines. Eclipse prediction? I’ve already discovered (likely rediscovered) simpler methods of predicting most lunar eclipses, and warning of solar eclipses. Nothing like Stonehenge is really required, neither its solstice alignments nor its 56 hole circle.

     Yet solstice alignments are clearly the most common feature of the hundreds of archaeoastronomy sites around the world, even if it is absurd to run a calendar keyed to them. What made them so popular? Religion might have spread solstice constructions around from one community to another, despite marginal practical utility. But in both Old World and New World? Surely we’re missing something here.

     Call the solstice alignments Exhibit B for the skeptical defense (whose chorus chants, “But you haven’t explained the archaeology yet!”). And I haven’t had any luck explaining Exhibit A either (the artificially smoothed horizons, such as the elevated bank at Avebury). Whatever were they using them for?