by at least 14% in recent decades.12 And in many ways, our
advanced technologies make us less flexible than previous so-
cieties in the face of change. We have made huge infrastructural
investments in coastal cities based on a bet that sea level will
remain constant. We have built sprawling cities in the desert
on the assumption that snow and rain will keep refilling reser-
voirs. We have a food production system that is predicated on
the belief that old, familiar weather patterns will always return.
But the weather is getting weird. Ten of the hottest years on
record have occurred since the start of this millennium. “‘One
hundred- year”’ and “500- year” flood events are happening once
a decade. The new rules of the Anthropocene are even making it
difficult for Earth scientists to use the quantitative models they
have developed to study geologic systems. Such models are
based on the concept of stationarity— the idea that natural sys-
tems vary within a well- defined range with unchanging upper
and lower bounds, an assumption that has yielded reasonable
predictions in the past. A sobering report by an international
group of leading hydrologists recently stated that “stationarity
134 Ch a pter 5
is dead and should no longer serve as a central, default assump-
tion in water- resource risk assessment and planning.”13 In other
words, the main prediction about weather and the water cycle
is that they will become increasingly unpredictable.
Yet the public clings to an optimistic belief in uniformitari-
anism. This is partly understandable, because it is rooted in the
geologic fact that climate in the Holocene Epoch, which saw
the rise of everything we associate with human civilization—
agriculture, written language, science, technology, govern-
ment, fine arts— has been exceptionally stable. In fact, this sta-
bility is arguably the very thing that allowed humans to build
civilizations at all. The large- amplitude climate oscillations
of the Pleistocene, in contrast, probably kept nascent human
societies in check. The “Ice Age” wasn’t, in fact, entirely icy;
instead, for 2.5 million years, the climate fluctuated manically
over many timescales— as glaciologist Richard Alley memo-
rably puts it, like someone “playing with a yoyo while bungee
jumping off a roller coaster.”14 Understanding what exactly was
going on in the Pleistocene is essential for putting current rates
of climate change in perspective. The story of deciphering the
Ice Age takes us back once again to Lyell, but also involves
Swiss farmers, a Scottish janitor, and a Serbian mathematician.
WA R M I N G U P T O I C E
Here in Wisconsin, large boulders of granite and gneiss are com-
mon centerpieces of upscale landscaping around medical com-
plexes and office buildings. In the early nineteenth century, such
stones— often completely different in composition from the
local bedrock— were among the most vexing mysteries facing
geologists in the Great Lakes states and northern Europe. These
erratics, scattered far from their sources, seemed to support the
Great acceler ations 135
biblical idea of a Great Flood. Consequently, the rocks, and the
clayey deposits they were often lodged in, became known as
diluvium (sediment left by the deluge) or drift (a rather gentle
term, considering the force of water that would have been nec-
essary to transport such material). The latter term persists,
anachronistically, in the name used for the distinctive region of
deep bedrock valleys in southwestern Wisconsin— the “Driftless
Area,” where no erratics or other types of diluvium are found.
A Swiss geologist, Louis Agassiz (1807– 1873), is usually
credited as the first to propose, in 1838, that great ice sheets, not
floodwaters, might have carried erratic boulders long distances.
Agassiz is championed in geology textbooks as a revolutionary
thinker, but it seems a German naturalist, Karl Schimper— who
in fact coined the term Eiszeit (“Ice Age”)— had earlier reached
the same conclusion and shared it with Agassiz on a joint outing
in the Alps.15 Schimper’s insights, in turn, may have come from
Swiss farmers, who understood glaciers and to whom it was ob-
vious that large boulders strewn far down alpine valleys marked
the former positions of ice masses. More unforgivably, Agassiz
later used his scientific credentials and his position as a Harvard
professor to advance completely unscientific and abhorrently
racist theories of human evolution; in my view, he should have
an asterisk next to his name in the annals of science, like an
athlete whose medal was rescinded for doping. Unfortunately,
he remains the eponym of a giant late- Pleistocene lake, glacial
Lake Agassiz, that covered much of North Dakota, Minnesota,
and Manitoba (and left them so famously flat).
While Charles Lyell disavowed the Flood, he also disliked the
idea of an Ice Age during which large areas of now- temperate
Europe and North America had been covered by ice. If not
exactly catastrophic, it was certainly non- uniformitarian. But
as geologists began to map the patterns of “drift,” the idea of
136 Ch a pter 5
a great Ice Age was shown to have explanatory power. In the
upper Great Lakes region, it became clear that there had in
fact been not one but several ice advances and retreats, each
leaving distinct deposits (though each, strangely, avoiding the
Driftless Area). What could be causing such cycles of warming
and cooling?
As early as the mid- nineteenth century, some scientists
began to explore the hypothesis that variations in Earth’s or-
bital habits could affect the way sunlight falls on the Earth
and potentially trigger episodic ice ages. The gravitational in-
fluences of the Moon and neighboring planets cause cyclical
changes in three aspects of Earth’s motion in space: (1) the
elliptical shape, or eccentricity, of the Earth’s orbit around the
Sun, which stretches and shrinks on a 100,000- year timescale;
(2) the tilt, or obliquity of Earth’s rotation axis, which varies
between about 21.5° and 24.5° every 41,000 years; and (3) the
slow wobble, or precession of the planet, like a toy top, which
changes the hemisphere that is pointed toward the Sun at the
solstices over a cycle that averages 23,000 years. Today, these
three variables are called Milankovitch cycles for the allitera-
tively named mathematician Milutin Milankovitch (1879–
1958) who, in spite of his status as a displaced person for most
of two world wars, managed to work out the combined effects
of these cycles on solar irradiance of Earth.
But it was actually a self- educated Scotsman, James Croll
(1821– 1890), who had performed the first arduous calculations
of the orbital cycles, more than 50 years earlier (a fact Milan-
kovitch fully acknowledged). Croll had a keen mathematical
mi
nd and great interest in science but was too poor to attend
even secondary school. After some years as an innkeeper, he
took a job as a janitor at Anderson College in Glasgow, where he
would study scientific volumes in the library late at night (in a
Great acceler ations 137
nineteenth- century real- life version of the plot of the 1997 film
Good Will Hunting). In the 1860s, he began a correspondence
with Charles Lyell about his calculations of orbital variations
and their effects on climate. Lyell, who by this time had reluc-
tantly accepted glacial theory, was impressed by Croll’s clear
brilliance and helped him gain a position at the Geological Sur-
vey of Scotland. (Croll also exchanged letters with Darwin on
the question of erosion rates). Croll’s work seemed to suggest
that ice ages would be out of synchrony in the Northern versus
Southern Hemispheres owing to the opposite effects of preces-
sion in the two regions. This reasoning appealed to Lyell, since
it meant that on average, the Earth maintained a steady state,
an idea Lyell could not relinquish. A half- century later, Milan-
kovitch would recognize that because of the disproportionate
concentration of landmasses in the Northern Hemisphere, the
influence of precessional cycles on northern latitudes actually
dominated global climate.
Neither Croll nor Milankovitch had any high- resolution geo-
logic data against which to test their calculations, however. By
the 1880s, the eminent Wisconsin geologist T. C. Chamberlin
(for whom my once- glaciated Svalbard valley was named) had
documented four distinct glacial periods, which he named for
the states in which their deposits were best preserved— starting
from the most recent, the Wisconsinan, Illinoisan, Kansan, and
Nebraskan. But there was no way to know the absolute ages of
these episodes nor whether there were still- earlier cycles of ice
growth and recession. The problem with land- based records
is that each ice advance, like a Zamboni resurfacing the ice
between periods in a hockey game, will tend to erode and over-
print the record of the previous events. Wisconsin (outside the
Driftless Area) was glaciated in all four ice advances, but it is
often difficult to recognize the deposits of the earlier three.
138 Ch a pter 5
In the last years of the nineteenth century, Chamberlin and
many others speculated about the causes of the Ice Ages, invok-
ing not only orbital cycles but also volcanism, mountain build-
ing, and ocean circulation. In 1896, Swedish chemist Svante
Arrhenius made the case that certain trace atmospheric gases,
particularly carbonic acid (H2CO3, carbon dioxide combined
with water vapor), could be important in governing climate be-
cause they are transparent to incoming short- wavelength radi-
ation (light) from the sun but block outgoing long- wavelength
energy (heat) reradiated from the Earth’s surface.16 (He even
surmised that emissions from coal burning might one day “im-
prove” Sweden’s climate). All these ideas would eventually
prove to be at least partly correct, but at the time, none could
be rigorously tested without higher- resolution information
about how climate had changed over time. There were many
climate suspects, but it was premature to bring them to trial;
the evidence was still too circumstantial.
E S P R I T D E C O R E S
Finally, in the 1970s, two new, rich archives of climate data that
revolutionized climate science were opened— as if someone
doing scholarly work with random volumes in a used bookstore
suddenly had access to the Library of Congress. These were
(1) deep- sea sediment cores obtained from a new generation
of oceanographic research vessels and (2) polar ice, sampled
through heroic international drilling operations in Antarctica
and Greenland. The deep seafloor and polar ice caps are simi-
lar in being sites where slow, continuous accumulation occurs
without interruption or disturbance, like dust gradually blan-
keting furniture in a closed- up room. Today, deep- sea cores
from all the world’s oceans provide a 160 million- year record
Great acceler ations 139
of global climate change (extending back long before the Ice
Age), encoded as variations in geochemistry and microscopic
fossils, at a resolution of thousands of years. Ice cores, in turn,
document 700,000 years of atmospheric variations that can be
read to the year, at least in young ice. Teasing climate informa-
tion from seafloor ooze and old snow, however, requires code-
breaking— translating the cryptic record of stable isotopes in
shell and ice.
Oxygen, like carbon, has two main stable isotopes, and in the
same way that light carbon (12C) is “preferred” by photosynthe-
sizing organisms over the heavier form (13C), light oxygen (16O)
is more likely to be taken up as water vapor during evaporation
than heavier 18O. This means that at any given time, precipita-
tion, including polar snows, will have less 18O and more 16O
than ocean water, and this sorting effect is further enhanced
during cold periods. During ice ages, when a significant fraction
of Earth’s water is locked up in glaciers and ice caps, the oceans,
and the organisms that form their shells from ocean water, will
have particularly high ratios of 18O to 16O. Conversely, glacial ice
will have particularly low values of this ratio. Ratios of ordinary
hydrogen (1H) to deuterium (2H) vary in a similar way, and so
in glacial ice (which is, after all, H2O) there is a second proxy
record of climate. Isotope ratios in sea sediments and ice thus
provide high- fidelity documentation of both global ice volume
and temperature over time.
The ice cores, and the much longer sea- sediment records,
reveal that Chamberlin’s four ice advances were just the most
recent of 30 that occurred over the 2.6 million- year span of the
Pleistocene. And the throbbing signal of the Croll- Milankovitch
cycles— a strong regular beat, with superimposed flutters— is
unmistakable.17 For the first 1.5 million years of the Pleistocene,
the 41,000- year obliquity cycle is especially evident. Then,
140 Ch a pter 5
around 1.2 million years ago, the pulse slows to the calmer
100,000- year rhythm of the eccentricity cycle, like an electro-
cardiogram readout for a patient who is falling off to sleep.
This is called the mid- Pleistocene transition, and its cause is
not completely understood. For one thing, of the three orbital
variables, eccentricity has the smallest effect on solar radia-
tion received by Earth, but for some reason the 100,000- year
cycle became amplified by geologic processes. There are also
higher- frequency “harmonics” in the climate records that do
not correlate with orbital variations. Recurrent temperature
oscillations of about 1500 years— the so- called Dansgaard-
Oeschger cycles— seem to be a characteristic internal rhythm
corresponding to the timescales of global ocean currents. This
means that the planet is not simply a puppet dancing to the
imposed rhythms of astronomical cycles but that it takes those
rhythms and riffs on them in its own way.
H E AT O F T H E M O M E N T
There is an even more important difference between the pre-
dicted effects of the combined orbital cycles and the observed
records of climate, and it further illustrates Earth’s capacity
to improvise on themes by Milankovitch. The Milankovitch
cycles are all essentially sine waves— symmetrical, palindromic
hills and valleys. When superimposed, they create more com-
plex patterns, but overall there is no systematic directional-
ity to them— at a glance, it wouldn’t be clear which way time
is flowing. The actual climate records from sea sediment and
ice, in contrast, have an asymmetric, sawtooth geometry: long
periods of cooling when Earth slid slowly into ice ages are
punctuated by short and abrupt episodes of warming. That is,
in each cycle, a tiny orbital nudge toward warmer conditions
Great acceler ations 141
was magnified by something in the Earth system into a heat
wave, like a thermostat gone haywire. The cause of this amplifi-
cation is preserved in the ice itself: greenhouse gases, especially
carbon dioxide and methane (CH4), or swamp gas.
When snow accumulates, air pockets remain between the
crystals (making snow shelters surprisingly warm, because they
are well insulated). At the poles, where snow doesn’t melt from
one season to the next, it compacts as it is buried, and at a depth
of about 60 m (200 ft) recrystallizes into ice. In the process, the
air pockets shrink, but vestiges remain as bubbles suspended in
the ice, like insects caught in amber. While there may be some
migration of air between layers in this process, the gas bubbles
trapped in polar ice are a direct record of past atmospheric
compositions at the resolution of at least decades. These tiny
bubbles tell us that over the last 700,000 years, global tempera-
tures have been correlated at the very highest level of statistical
significance with the concentrations of the greenhouse gases
carbon dioxide and methane.
So how could greenhouse gases take a small increment of
Milankovitch warming and magnify it into a meltdown? The
Timefulness Page 16