Ptolemy's Almagest dominated astronomical thought from when it was written in the 2nd century CE, until the 16th century. Yet, as historically important as this work is, its nearly impossible to find out anything beyond a surface level understanding of what it contained.
I've spent nearly 7 years working through this text, making sure I understand every calculation. And now, I'll explain exactly what this book contains. In this series, I'll walk through the text - first explaining it conceptually, and then exploring select mathematical topics in future videos.
This video covers Book IV in which Ptolemy derives the first lunar model.
Written in the 2nd century CE, Claudius Ptolemy's Mathematicae Syntaxis, better known as the Almagest,
became the most important text in mathematical astronomy for the next 1500 years. But despite
the historical significance of this text, it's nearly impossible to find detailed information
on how its models were derived and functioned. So much so that when I began studying historical
astronomy, I decided it was best to sit down and work through it myself. So in this series,
I'll share what I've learned about one of the most important astronomical texts of all time.
I'll begin by addressing things conceptually, so viewers that aren't interested in the detailed
mathematical workings can get an appreciation for Ptolemy's work. With that out of the way,
I'll then explore select mathematical topics, taking the understanding from the conceptual
to functional, allowing viewers to understand how models were practically applied.
In this video, I'll be exploring Book 4. In this one, Ptolemy begins exploring the lunar model,
developing his first version, believed to be based on the model of Hipparchus.
So join me as we learn how Ptolemy defined astronomy before the telescope.
Ptolemy begins Book 4 by asking the question of what sorts of observations will be needed
to calibrate the lunar model. In Book 3, Ptolemy got away with determining the position of the sun
for calibrating his solar model by a trick of geometry. By definition, the sun is a quarter
of the way through its motion about the zodiac at each successive equinox and solstice, allowing
its position to be precisely defined so long as the dates and times of those events could
be determined. For the moon, he will exploit another geometric necessity by making use of
eclipses. When an eclipse occurs, either solar or lunar, we know that the moon must either be
directly in front of the sun for a solar eclipse or directly opposite the sun for a lunar eclipse.
Since we now know how to calculate the position of the sun at any given time,
that means that we will be able to precisely know the position of the moon as well, at least
insofar as the solar model is accurate. But there's a catch. The moon is close enough
that it exhibits significant parallax. This actually means that solar eclipses aren't good
because our position on a spherical earth can allow for a solar eclipse even when the
moon's center is not precisely aligned with the ecliptic. Thus, Ptolemy will only make use of
total lunar eclipses, which play out independently of the observer's position on earth.
You may be concerned because the width of the earth's shadow at the lunar distance is about
2.5 times the moon's diameter, which means that it may be slightly above or below the ecliptic.
But Ptolemy, as we'll see, is primarily concerned, for now, with the ecliptic longitude,
which, at mid-eclipse, is pretty close to the straight line Ptolemy seeks.
Ptolemy then goes on to give a brief description of how lunar eclipses work,
noting that the earth casts a cone-shaped shadow behind it, and that lunar eclipses are the moon
passing through this shadow. And while solar eclipses play out differently depending on your
location, lunar eclipses are identical to every one on earth, with the exception of happening
at different local times. Just like Ptolemy began Book 3 with a discussion of the period of the sun,
he now goes into what a period of the moon means, and it's significantly more complicated.
As Ptolemy tells us, the moon's motion appears anomalistic both in longitude and latitude.
The time it takes to traverse the ecliptic is not constant, and neither is the time that it
takes to return to the same latitude. What Ptolemy is telling us here is that the moon's motion is
complicated. The sun is, by definition, confined to the ecliptic, although it moves around it at
a somewhat irregular rate, speeding up and slowing down as it goes around. But the moon
has its own circle that it appears to trace out in the sky, which is inclined through the ecliptic
by about 5 degrees. This creates a bit of a problem, since the ecliptic is our fundamental
plane that we measure by. In general, the moon follows the ecliptic, but since its circle is
slightly inclined, it will bounce up and down relative to the ecliptic. However, when the moon
returns to the same point in longitude, it won't necessarily be at the same latitude that it
started at. Its cycles are out of sync. So how do we deal with this? How can we even define a lunar
period? The answer is that there isn't one answer. We'll need to consider and come up with average
periods for all of the types of lunar periods that the moon can exhibit. Defining those periods,
and putting together a plan on determining them, is the real focus of Chapter 2.
Even if the time that it takes for the moon to return to the same ecliptic longitude isn't the
same as the time it takes to return to the same ecliptic latitude, if you wait long enough,
eventually you'll have a return in both latitude and longitude at the same time. It's kind of like
watching the turn signals on two cars in front of you. They go out of sync and then come back
together. Their periods are different, but after a while, they'll briefly line up again,
before getting out of sync and then lining up again. Ptolemy states that the ancient astronomers,
likely referring to the Babylonians, looked for cycles in the moon's behavior by looking for
patterns in eclipses, and they seem to have found one with a period of 6,585 and a third days,
just over 18 years. Today, we call this period the Saros Cycle. In that period of time,
the moon seems to go through an integer number of returns in latitude, anomaly, and lunar phases.
Let's first consider the lunar phases that I just mentioned. In a Saros Cycle,
the moon goes through a full cycle of phases known as lunations 223 times.
If you divide 6,585 and a third by 223, you would get the mean period for a lunation.
This is known as a synodic month. Second, as we noted, the moon has anomalistic motion,
in which it speeds up and slows down. The amount by which it does so is fairly dramatic.
Some days the moon moves by about 10 degrees, but others it moves as much as 14 degrees in a single
day. For scale, the moon's angular diameter is about half a degree, which means that that
difference of 4 degrees is about 8 times the width of the full moon. In 6,585 and a third days,
the pattern of anomaly will occur 239 times, and again if we divide 6,585 and a third by 239,
we get the average period of anomaly. This is referred to as an anomalistic month.
Third, if we track the moon bouncing up and down with respect to the ecliptic,
we'll find that in a Saros Cycle, it does so 242 times. And again, we can divide the Saros
period by this to find the average duration of a return in latitude. This is called the draconitic
or draconic month. Lastly, we can ask how long it takes for the moon to return to the same
ecliptic longitude. However, the moon doesn't actually have an integer number of returns in
ecliptic longitude in a Saros Cycle. The reason should be pretty obvious if we think about it
for just a minute. It's because the Saros Cycle incorporates an integer number of lunar phases,
which happen when the moon returns to the same position relative to the sun. However,
since the Saros Cycle isn't an integer number of years, the sun won't have returned to the same
ecliptic longitude in that amount of time, and thus neither will have the moon. But Tony still
wants to determine an average period for the moon to complete a circuit about the ecliptic.
Fortunately, we can still get there using the Saros Cycle. We're just going to have
to do a bit more work than dividing the Saros period by the number of returns.
Instead, we need to notice that, in a Saros Cycle, the sun has moved 10 and two-thirds degrees
further than an integer number of 360-degree revolutions around the ecliptic. Since the
Saros period returns the moon to the same position relative to the sun,
the moon too has moved by that much extra. Ptolemy then notes that some people don't
like the fact that the Saros Cycle has that lingering third of a day in it. I guess
fractions are hard or something. Fortunately, if you want to get rid of it, you can multiply the
Saros period by three to get a related cycle, referred to as the Exolygmos, which is 19,756
days long. I should note that I've been doing a bit of foreshadowing in this chapter.
Although I've been describing how we could get an average period for each of the types of lunar
months, Ptolemy holds off on the actual calculations until the next chapter. That's
because the vaunted Saros Cycle we've been discussing was already known to be wrong,
thanks to Hipparchus. Instead, the period Hipparchus uses is 126,007 days and one hour.
In this period, he finds 4,267 lunations, 4,573 returns in anomaly, and 4,612 returns in longitude
minus 7.5 degrees, again related to the sun's position being different.
And because this cycle has a longer baseline, dividing the total period by the number of
returns to figure out the average lengths of the various months should give an even more
accurate determination for each of the types of lunar months. But if you're paying close attention,
you may have noticed that I didn't give a number of returns in latitude for this
Hipparchan Cycle. Ptolemy calls this out as well, noting that this Hipparchan Cycle
didn't have an integer number of returns in latitude. It was very close, and the eclipses
still repeat, but the size and type of the obscuration is slightly different, meaning that
the pattern isn't perfect. This is frustrating because it means that we can't use this cycle
to determine the length of a draconic month, at least not directly. Instead, Ptolemy cites
Hipparchus again, stating that there were 5,458 lunations for every 5,923 returns in latitude.
Thus, by knowing the synodic month accurately, we can use this ratio to determine the missing
draconitic month. If you're feeling lost with all of this, Ptolemy certainly shared your pain
and acknowledged the complexity, stating, it is not simple or easy to carry out, but demands a
great deal of extraordinary care. Finding a true cycle of eclipses is challenging because it's
possible to get false positives where you may think that you've found a cycle in which a few
eclipses happen in a similar fashion in a similar amount of time, but this could just be a
coincidence. He then goes on to describe the necessary conditions for establishing a true
cycle of eclipses. Yet, despite all the careful consideration that Ptolemy tells us Hipparchus did,
he still determines that there are issues with the periods of anomaly and latitude,
so considerable as to become quite apparent to us.
However, he doesn't address them here and promises corrections which we shall derive later.
In chapter 3, Ptolemy completes the calculations to determine the mean periods as we described
previously. However, instead of expressing them as periods of return, he expresses them in degrees
per day. And, as we did with the sun, we'll also be making a lunar mean motions table for convenient
use later on. However, in addition to the mean motions in longitude, anomaly, and latitude,
he also takes the difference between mean motions in lunar and solar longitudes.
This will be used to determine how far away the moon is from the sun about the ecliptic,
a property known as elongation. This is important in determining lunar phases for eclipses which
we'll be exploring in book 5. The tables themselves are in chapter 4 and broken into
the 18-year periods, single-year periods, Egyptian months of 30 days each, single days, and hours,
just as they were in the solar table of mean motions. However, due to the issues Ptolemy
hinted at at the end of chapter 2, the values for the periods for anomaly and latitude don't
actually correspond to what we would get if we performed the calculations based on the periods
we just discussed. Rather, Ptolemy has sneakily used values based on corrections to the anomaly
and latitude which he has yet to discuss and which we'll come back to in chapter 7 for the anomaly
and chapter 9 for the latitude. With these mean motion tables laid out, Ptolemy is ready to begin
the real work on the lunar model, again asking which hypothesis should be used, the eccentric
or epicyclic. And again, he concludes that it doesn't matter for the simple lunar model,
which sure sounds like it implies a more complicated one. Spoiler, there is,
but we'll come to that in due time. Ptolemy almost immediately lays out why we can use
either lunar model for now. He states that the model we're about to develop is treated as if it
only has one anomaly in longitude, and that this anomaly corresponds to the above period of return,
which is the only one which our predecessors, just about all of them, have hit upon. But,
as we'll see later, Ptolemy discovered a second anomaly linked to its distance from the sun,
i.e. the elongation. He tells us that this initial model works well when the moon is precisely at
conjunction and opposition, full or new phase, but has problems near the first and third quarter
phases. It shouldn't be too surprising that this model works well for the full moon since we're
going to be building it using lunar eclipses which happen at just this phase. Ultimately,
Ptolemy settles on using the epicyclic hypothesis for this lunar model. To calibrate it, he begins
by trying to figure out the size of the smaller circle, the epicycle, as a proportion of the
larger circle, the deferent. As he did with the sun, he doesn't try to determine the sizes in
absolute terms. Instead, he again treats the deferent as a unit circle with a radius of 60
parts, and then gives the size of the epicycle, again in parts, in relation to this. Obviously,
since both the sun's eccentric circle and the moon's deferent circle both have radiuses of 60
parts, the parts for each of these are different size parts. Confusing to have two different units
with the same name, I agree, but this is quite common, and Ptolemy will be bringing them into
a single system of measurement later on. But to determine the sizes of these circles,
he uses a series of three eclipses, which I'll call A, B, and C, which came from the Babylonian
records. He first determines the ecliptic longitude for each. This is done by calculating
the position of the sun for each of them using the solar model. Then he can add or subtract 180
degrees to the solar longitudes to determine the lunar longitudes, since the moon must be opposite
the sun. Next he pairs off the eclipses, A with B and then B with C, and asks how far the moon
moved between each pair. However, this total motion is apparent motion, which is comprised
of both the mean motion, about the deferent, and the anomalistic motion, caused by the motion
about the epicycle. Fortunately, we can determine how much of that motion was caused by the motion
about the deferent by taking the interval of time between the members of each pair,
and then using our lunar mean motion table. Then whatever's left must have been caused by
the epicycle. Essentially what Ptolemy has done is subtracted out the mean motion from the deferent.
This allows him to only consider the motion about the epicycle. Then he can sketch out the
residual angles that were caused by the epicycle from the vantage point of Earth. At this point,
we don't know if the moon was on the near side or the far side of the epicycle for each eclipse,
but doing this will cut off some arcs on the epicycle. Ptolemy is able to determine how long
the arcs should be by taking the interval of time between the eclipses in each pair,
and then determining how far about the epicycle the moon should have moved from the mean motion
table. From this, he determines that arc AB should be 306 degrees and 25 minutes,
and arc BC should be 150 degrees and 26 minutes. Then Ptolemy can ask how he must arrange the
epicycle and the eclipses on it so that everything lines up. This reveals that the eclipses must have
been arranged like this, with A and B on the far side and C on the near side. Then, through some
geometry, he's able to determine the distance at which the epicycle needs to be positioned
as well as its radius. To check himself, he repeats the same calculations using a series
of three eclipses observed in his own time to show that the size of the epicycle, based on
his contemporary observations, is in nearly perfect agreement with that found from the
Babylonian eclipses. As I promised before, Ptolemy now discusses
the correction that he did regarding the mean period of anomaly. Now that we have some hard
numbers for the lunar motion instead of the somewhat fuzzy logic from before of well these
periods all seem to line up after a Hipparchan cycle, Ptolemy has checked the values to ensure
that they match with what's expected. The mean period for longitude seems good, but the mean
period for anomaly is off by about a quarter of a degree. Thus, when computing the mean motions for
the table in chapter 4, they have been computed with this minor adjustment already built in.
Now that Ptolemy has gotten the base layout of the lunar model all set up,
we'll again need to determine an epoch position from which to perform our calculations.
This is done by picking one of the eclipses for which we just determined the lunar position.
We can then use our tables of mean motion to rewind the motion about the epicycle and the
deferent, and figure out the orientation at the beginning of the epoch. For that epoch date,
Ptolemy uses the same as he did for the Sun, the first day of the reign of Nabanassar,
the 26th of February 747 BCE. Doing so, he finds that the Moon's mean position was 11°22' into
Taurus at that time, which was 70°37' from the mean position of the Sun. Then, for the position
on the epicycle, he found that it was 268°49' from Apogee. But notably absent from the list
of values for the lunar epoch is where the Moon was in ecliptic latitude at that time.
And this is because he still needs to discuss the correction for the mean motion in latitude.
He does so in the next chapter. There, Ptolemy tells us that there is an error because Hipparchus
had some parameters of the Moon incorrect, and Ptolemy says that better methods were available
in his time. So he proposes to use them. Ptolemy again searches through the ancient records to
a pair of eclipses, but this time with very specific properties. What he was looking for
were eclipses that were very similar in nature, being not quite total, but obscured by the same
amount on the same side for the same amount of time, and while the Moon was roughly at the same
distance from the Earth. He imposes these criteria because it means that the Moon must have been at
approximately the same ecliptic latitude for each eclipse, which we can tell because the same
amount of the Moon was outside of the umbral shadow, which has a pretty consistent diameter.
Lastly, he required that these eclipses both happen near the same node.
Finding a pair of eclipses that suits these requirements, he is able to geometrically work
out how far the Moon moved about its epicycle in that time by using the mean motion tables,
and he notes that the Moon wasn't in the same position about the epicycle at each of those times.
This indicates that the Moon hadn't completed a full cycle in longitude, which should also be
expected for the pair of eclipses since Ptolemy specified that they both occur near the same node.
In other words, Ptolemy has demonstrated that the Moon returned to the same ecliptic latitude
faster than should have been expected, and knowing this, he's able to work up a minor
correction which he uses for the values in the mean motion tables. And with that improved value,
we can then determine the position of the Moon in ecliptic latitude at the start of the epoch,
if we know the lunar latitude at a specific point in time that we can rewind from.
You might be tempted to think that we could use basically any of the eclipses that we've
just been talking about, since for eclipses to occur, the Moon must be close to the ecliptic,
but it need not be as close as you might expect. Again, the Earth's shadow is 2.5 times the width
of the full Moon. This means that the Moon could have been a fair distance above or below the
ecliptic, and you'd still have a total eclipse. Thus, we don't know its true location. However,
the further it is from the ecliptic, the shorter the eclipse's duration. So Ptolemy could have done
pretty well just by choosing the total eclipse with the longest duration, and been pretty sure
it was very close to the ecliptic. But instead, he chooses to do something more complicated.
He again finds a pair of eclipses with some specific requirements. This time, he wants ones
which have the mean positions of the Moon roughly the same distance from opposite nodes. Using some
geometry, he is able to determine how far the Moon's mean position was from the northern point,
the point at which the Moon has its greatest latitude, for one of the eclipses. Then,
it's a simple matter of rewinding the Moon from that position by the amount of time between that
eclipse and the epoch. Doing so, he finds that the Moon's mean position was 354°15' from the
northern point at the start of the epoch. Having concluded this, Ptolemy then tells us that we're
ready to create our table of anomaly, which will tell us how far ahead of, or behind, the mean
point that the Moon is thanks to its epicycle, based on how far around the epicycle it is from
Apogee. This is then laid out in Chapter 10, wherein the first two columns are the distance
of the Moon about the epicycle, and the third is the amount of anomaly. If the Moon is 0°180'
from Apogee in the first column, then the anomaly should be subtracted from the mean position,
and if 180°-360°, it gets added. That basically completes the lunar model for now.
However, Ptolemy puts in one more chapter, which strikes me as more of a historical footnote than
really part of the model itself. Specifically, he notes that Hipparchus evidently had two values
for the maximum lunar anomaly. In one case, he states that it was 5° in 49 minutes,
whereas in another, he gives it at 4° in 34 minutes. So how did he arrive at two different
numbers for the same value? Ptolemy states that this is because Hipparchus derived lunar models
using both the eccentric hypothesis and the epicyclic one separately, resulting in a different
maximum anomaly for each. This should surprise you since we previously stated that the eccentric
and epicyclic models should produce the same results, but that was only true if certain ratios
related to the eccentricity were kept consistent between the two models. Ptolemy reviews how
Hipparchus worked out the values and finds that Hipparchus did not have those ratios the same,
which explains the discrepancy. But why didn't Hipparchus keep those ratios the same? Ptolemy
attributes it to an error in the math in which Hipparchus incorrectly determined the length of
time between two of his eclipses. Ptolemy then recalculates the base parameters for the eclipses,
demonstrating where Hipparchus's errors likely were, although he stops short of recalculating
the eccentricity specifically. I found the choice to stop the calculations there a bit of an
interesting one as, without doing so, he doesn't demonstrate that he can truly reproduce Hipparchus's
values. Regardless, that digression closes out book 4. At this point, we have a working lunar
model which allows us to calculate the lunar position at any time. This model is almost
certainly similar to the one that Hipparchus created, at least in how it's constructed with
the moon on a simple epicycle, although, as we've seen, Ptolemy reviewed and tweaked specific
parameters. But Ptolemy has also told us that it's fundamentally flawed and doesn't produce
accurate results near first and third quarter. So, in book 5, Ptolemy is going to add a new feature
to the model to help account for this, but we'll save that for next time, as we continue to explore
Astronomy Before the Telescope.