In winter 1634, James Byron “Jabe” McDougal “recalled that “the winter of 1631 – 32 had been quite a shock to himself and his fellow up-timers. Not only were they considerably farther north than they had been when Grantville was in West Virginia, but they were smack in the middle of what up-time historians had called the “Little Ice Age,” which had begun some two centuries prior and would continue for another century, give or take. (Robinson, “Mightier than the Sword,” Grantville Gazette 6).


So what was this Little Ice Age? The real Ice Ages were prolonged (as in millennia) periods of pronouncedly colder world or hemispheric temperatures in which the polar and continental ice sheets were of considerably greater extent than in historical times. There have been a dozen or so major glaciations over the last million years. A particularly big one occurred 650,000 years ago and lasted 50,000 years. However, the one that is usually considered the last Ice Age peaked about 20,000 years ago.

So that implies that a little ice age is one that is shorter and milder than that one, yet still noteworthy. Defining when a little ice age begins and ends is a bit tricky. Do you draw the line based on when a particular glacier advances or retreats, when a particular lake freezes or thaws, or when the grapes are harvested? If you rely on mean temperatures, then over how many years do you average them, and to what longer reference period do you compare that moving average? What if the temperature “breakpoints” are different in Iceland than they are in France?

Depending on who you ask, the Little Ice Age began in 1200, 1250, 1300, 1350, 1400, 1450, 1500, 1550, 1600 or even 1650. There is more agreement as to when it ended; 1850 is the year usually cited, but some would say 1870, 1900, or even 1920.

When people talk about the Little Ice Age (LIA) nowadays, they are mostly interested in the Big Picture: Was the LIA, viewed on some appropriate time scale, a global, a hemispheric or merely a European phenomenon? How much colder was the earth then than it is now? What caused it? Is the Earth warmer now than it was during the “Medieval Warm Period” that preceded the LIA? And to what extent is that warmth attributable to human activity (changes in albedo as a result of deforestation, or increases in greenhouse gases as a result of factory emissions)?

However, for those writing in the 1632 Universe, the Little Picture is what we need: what is the climate likely to be like in Germany, Italy, France, Scandinavia, England and in other areas of interest, each year over the decade following the Ring of Fire (RoF)? (The RoF occurred up-time on April 2, 2000 and down-time on May 25, 1631 Gregorian calendar.) How does it compare to the climate in living memory, for both the down-timers and up-timers? What practical effect will it have on health, agriculture, transportation, communications, mining, industry and warfare?

In the first part of this article, I will provide some background as to the effects that climate can have on human society.

In the second part, I will try to fill in the Little Picture, based on the assumption that the Ring of Fire has not altered world climate; i.e., I can rely on modern reconstructions of historical temperature and precipitation averages in the areas and years of interest.

Finally, in the third part, I will consider the Ring of Fire as a meteorological phenomenon, and speculate about how much and for how long it could perturb weather and even climate.


Climate and Health

Excessive heat and cold can directly threaten human life. In studied regions of England and Wales (1993 – 2003 data), it was found that risk of mortality increased by 3% for every degree Celsius above the “heat threshold” (95th percentile of the mean daily temperature for the region), and by 6% for every degree below the “cold threshold” (the 5th percentile). In general, heat effects were seen once mean temperature reached 17-18oC, and cold effect below 5oC. (Hajat). At least in modern Europe and the United States, cold-related deaths are more common than heat-related ones, and that was even more likely to be true in LIA Europe.

The very old and very young, and those in poor health, are the most vulnerable to temperature extremes. However, the human body can adapt over time, which is why we can live in both cold and hot climates.

In addition, there are “cultural” as well as biological adaptations, and these can work in the short-term. In cold weather, one can wear heavier clothing, or go indoors and build a fire. In the 17th century, there was less that could be done about hot weather, of course. Especially since many Europeans thought that bathing was a bad idea.

Lives may be also be lost as a result of flooding caused by excessive rainfall, if the endangered population cannot flee to higher ground in time. Drought can also kill, if water has not been stored in advance. In hot, dry climates, dehydration is often associated with heat stress.

Even when climate extremes don’t kill you outright, they can cause famine, which in turn reduces the body’s resistance to infectious disease. “Malnutrition aggravated an influensa epidemic of 1557 – 8” (Mandia).

Normal seasonal variations may also have health consequences. Sometime around 400 B.C., Hippocrates declared, “The changes of the season mostly engender diseases.” The basis for seasonality is not always clear. It may be related to increased pathogen (or disease vector) survival under particular temperature and humidity conditions, increased opportunity for transmission as a result of travel or overcrowding, or reduced host immunity or impairment of other host defenses (e.g., drying of the mucous membrane).

That said, some diseases definitely have seasonal propensities. In autumn and winter, we have influenza; in spring, measles; in summer, malaria (and in modern times, polio). (Dowell). In 1908 Manhattan, scarlet fever and measles were most common in March; there was a higher incidence of death from pneumonia and bronchitis from November through April; death from childhood diarrhea peaked in July – August, and cases of typhoid in August – September (North).

For malaria, the role of climate is well-understood. “Malaria transmission does not occur at temperatures below 16oC or above 33oC, and at altitudes > 2000m because development in the mosquito (sporogony) cannot take place. The optimum conditions for transmission are high humidity and ambient temperature between 20 and 30oC. Although rainfall provides breeding sites for mosquitoes, excessive rainfall may wash away mosquito larvae and pupae.” (Cook 1202). The northern limit for malaria in Europe has been the 15oC July isotherm (Reiter).

While Europe was colder during the LIA, it wasn’t cold enough to prevent malaria. However, a correlation has been reported between high (over 16oC) summer temperatures in Kent and Essex parishes; Reiter speculates that the “hot weather . . . could certainly have increased the probability of transmission by shortening the extrinsic incubation period (the time required for the mosquito to become infective after feeding on an infected person).”

Yellow fever also is seasonal. In Trinidad, the density of one mosquito carrier was six times more common in the wet season (May – November) than in the dry season; bear in mind that in the tropics; the seasonal variation of temperature is small (Chadee).

Plague is rather more problematic. In Switzerland, 1628 – 30, the outbreaks were mostly between September and January, with November the month of highest frequency (Eckert). But other outbreaks favored summer, with peaks of mid-summer for Penrith 1597 – 8, Marseilles 1720, and London 1665, and late-summer for London 1625 and Debrecen 1739 (Welford).


The climatic deterioration was blamed on human misconduct. In Switzerland, Cysat wrote in 1600, “Unfortunately because of our sins, for already some time now the years have shown themselves to be more rigorous and severe than in the earlier past . . . .” (Pfister2007).

From blaming sins, it was a short step to looking for sinners. In Treves, Hans Linden’s Gesta Treverorum blames the nigh-continuous crop failure of 1581 – 99 on “witches of devilish hate,” and proclaims that “the whole country stood up for their eradication.”

Accusations of causing “unnatural weather” or crop failure peaked when climate extremes disrupted agriculture. Moreover, it was generally considered unlikely that a single witch could control weather on a large-scale, which meant that the witch hunts were comparably large in scale (Pfister2007, Behringer).

In the 1620s, in Central Europe, there was a succession of extremely cold summers. For example, on May 24, 1626, there was a hailstorm in Stuttgart, “which brought hailstones the size of walnuts . . . .” Two nights later, ice formed, and crops failed. Witch-burnings in central Europe rose to a peak of over 500 a year, well above the “normal” (presumably, non-weather-related) level of the mid-16th century of 100 a year. As late as 1630, “suspects still had to confess that they had been responsible for the extreme frost in May of 1626.”


The wealthy, of course, get to choose where they live, and they live where conditions are healthiest. Lamb has pointed out that in Surrey, 20th-century luxury housing is on the hilltops, whereas in the LIA, the favored sites were in the valley bottoms (LambCHMW 251).


Climate, Agriculture and Fishing

Jabe McDougal was not the only up-timer who has the Little Ice Age on his mind. One May, after the death of Mabel Jenkins in 1632 (Grid), Joe Jenkins grumbles that “there’s snow on the ground” and “it’s still here from February.” He is worried that it won’t be gone in time to plant corn and tomato, and adds, “If it weren’t for the wheat, I could just up and starve with this here ‘Little Ice Age.'” (Howard, “Golden Corn—A Tale of Old Joe on the Mountain Top,” Grantville Gazette 9).

In the broader scheme of things, climate change can affect what crops can be raised in a particular part of the world. The ability of a plant to grow in a particular place is dependent on soil and climate.

Too much or too little heat, or too much or too little rain, can result in crop failure. Both droughts and floods can kill crops. Floods can be caused, not just by excessive rain, but by normal rain after a prolonged dry spell, as a result of which the soil has lost its normal ability to absorb water (Brooks 60).

If food cannot be rapidly and economically brought in from an unaffected area, crop failure leads to famine. Famine several years in a row can result in a major increase in illness, death or emigration, or in political unrest resulting in overthrow of the government or bloody suppression of a rebellion.

The down-timers in Thuringia are growing grain (primarily rye, barley, and spelt), vegetables, grasses for hay, and woad for dyeing. Of course, those are already adapted to the local climate. How will the plants that passed through the Ring of Fire, and are accustomed to the conditions of West Virginia in 2000, fare in LIA Germany?

There are complex plant-specific crop models available for predicting the combined, nonlinear effects of temperature and rainfall on plant development. These take into account changes in the sensitivity of the plant depending on its growth stage.

That’s too complex for us, but we can look at what are called “cardinal temperatures”—minimum (base), optimum, and maximum (ceiling). Even those have their subtleties, as the cardinal temperatures may differ for germination, vegetative growth, and reproductive yield (which for grains is the crop yield).

Generally speaking, cool season crops (oats, rye, wheat, barley) have a base of 0-5oC, an optimum of 25-31oC, and a ceiling of 31-37oC, and hot season crops (melons, sorghum) have a base of 15-18oC, an optimum of 31-37oC and a ceiling of 44-50oC (Change, Climate and Agriculture 75).

Crop maturation is a cumulative process and crop scientists sometimes use the concept of growing degree days, awarding one GDD (oF or oC) for each degree (oF or oC) that the mean temperature on a particular day exceeds the base (some versions truncate if the temperature exceeds a ceiling). For example, wheat has a base of 40oF; corn, 50oF; and cotton, 60oF. Insects also have GDDs; 50oF for the European corn borer. (Fraise).

A decline in mean summer temperature has a double whammy. It reduces both the height and breadth (growing season length) of the GDD curve. In England, in the coldest years of the LIA (1695, 1725, 1740, 1816), summer temperatures were about 2oC below the modern norm, and the growing season “was probably shortened by two months or even more.” (LambCHMW 223).

The principal Indian crop in New England was maize, and there’s reason to believe that the native strains required 2000 growing degree-days (GDDs), base 50oF, to reach maturity. (The Indians also grew beans but these reached maturity more quickly.) In the 1960s, Connecticut, Rhode Island, Massachusetts, the Connecticut River Valley (NH-VT border), southeast New Hampshire and southwest Maine all were receiving at least 2000 GDDs (the area around Boston typically received over 2500 GDDs). A 2oF reduction in mean July and mean annual temperatures would put all of New Hampshire, Vermont and Maine, as well as northwest Massachusetts, under the 2000 GDD mark (Demeritt).

Grantville is based on Mannington, located in Marion County, WV. According to the 1997 Census of Agriculture, Marion County had only one farm growing wheat and oats for grain. It had 251 farms producing hay (primarily from alfalfa). You can figure that alfalfa would be cut at 750 GDD, base 41oF, to yield a fiber content 40% neutral detergent fiber. For 45% NDF, you would allow another 220 GDD (Pennington).

While there is no commercial production of corn in Mannington, canon says that there was a small quantity of seed corn available in Grantville as of the RoF (Weber, “In the Navy”, Ring of Fire 1). There are also sunflower seeds, see Vance, “Second Chance Bird, Episode Two,” Grantville Gazette 33. Sunflowers have a base of 44oF and require a GDD of something like 2300. (Putnam).

We can compare these temperatures to those that are reconstructed for the places and times of interest.

Bear in mind that temperatures below the base temperature might not just stop growth, they might kill the plant altogether. Flowers and young fruits of fruit trees are often killed by mild frosts (0-5oC) (Hatfield).

The USDA defines plant hardiness zones based on the extreme cold (expressed as the average minimum annual temperature) that a particular plant can tolerate. Zone 1 is -60oF to -50oF, zone 2 -50 to -40, and so on up to zone 11, 40 to 50. Each zone may be further subdivided into two subzones, “a” and “b,” with “a” as the colder half. (Zones 0 and 12 are special cases; 0a is under -65oF, 0b is -65 to -60, 12a is 50-55 and 12b is over 55.)

Plants vary in terms of what kind of climate they like. For example, the orange tree (Citrus sinensis) is considered hardy in zones 9a – 11a, whereas the Scots pine (Pinus sylvestrus) grows in zones 1-4.

In 1990, the USDA prepared a map of North America depicting which areas are in which hardiness zones, based on their average annual lows (over the period 1974 – 86). This of course changes as the climate changes; in 2006, the Arbor Day Foundation updated the U.S. hardiness zones to reflect the most recent 15 years of data and perhaps half the U.S. (excepting California and Nevada) experienced a one zone (10oF) increase.

The logic behind the hardiness zone definition is that even a brief exposure to a cold enough temperature will kill the plant. However, it ignores the fact that a plant may withstand a short exposure to say -5oC yet be killed if there are too many days at 0oC.

Also, it ignores the effects of day length, summer heat, wind, and the amount and distribution of rainfall, which in turn are influenced by latitude, elevation, continental position, and mountain barriers. The American Horticultural Society has a Plant Heat Zone map; the zones are based on the average number of days per year above 30oC, thus accounting for summer heat. There are other zoning systems, that take additional factors into account, but we can’t use them in LIA Europe because we lack some of the necessary data.


Besides the direct effects of climate on plant growth, there are also indirect effects. Plant pests are also affected by temperature; a warm winter may mean a bumper crop of insects in the spring. In late 17th-century Switzerland, cool springs led to crop losses as a result of attacks of the parasite Fusarium nivale, which is active under snow cover (LambCHMW 206).


Domestic animals are also affected by climate. Animals can be killed by climate extremes, especially the combination of heat and drought. Even conditions that don’t kill can reduce reproduction, growth rate, and milk production.


Considering domesticated plants and animals together, both temperature and precipitation can have significant adverse effects. The so-called LIA-type impacts are:

March, April: cold decreases forage for dairy animals and the volume of the grain harvest.

July, August: rain interferes with the harvesting of crops.

September, October: cold forces animals into the barn earlier and reduces the sugar content of vine-must; prolonged rain reduces area sown and nitrogen content of the soil (thus affecting the following year’s productivity).

Pfister2006 has combined temperature and precipitation monthly data to arrive at a “biophysical climate impact factor.”


Fish have preferred water temperatures. During the LIA, cod and herring moved south, hurting the fisheries of Norway, Scotland and the Faeroe Islands, but benefiting the English (Mandia).

Climate and Transportation

In 1630, the cheapest form of transportation was by water. However, except in far northern Europe, transport was dependent on liquid water; skating or skiing on ice or snow was fine for individuals but not practical for large-scale freight movement.

So that means that we need to ask when will geographically significant navigable rivers freeze and thaw, in which months will strategic harbors be closed by sea ice, and when will particular sea routes be endangered by icebergs.

Rainfall can also make a difference. In some parts of the world, rivers are navigable only for part of the year. Or in some years and not others.

Land transportation is also affected by climate. Snow can close a mountain pass, or simply make it slower to travel by road. Rainfall can turn dirt into mud, or make a ford impassable, or cause a flood that destroys a bridge.

On the other hand, the freezing of rivers (while not good for water travel) can make river crossings easier. In 1597-8, Matteo Ricci wrote that “once winter sets in, all the rivers in northern China are frozen over so hard that navigation on them is impossible and a wagon may pass over them.” (Brooks 55).

Up-time transportation technology also has its vulnerabilities. Cold temperatures can reduce starter battery life, render fuel viscous, and cause engines to stutter. High temperatures make it easier for engines to overheat.

Climatic interference with transportation can make it more difficult to relieve a local famine by moving in food from elsewhere.

Climate and Communication

Prior to the RoF, messages traveled, at least over distances beyond line-of-sight, at pretty much the same speed as people and goods. On land, the fastest communications were those provided by a post horse system, and at sea, messages could be carried by a sailing ship built for speed and not burdened with a heavy cargo. The effect of climate on these channels of communication have already been discussed in the context of transportation.

The up-timers will be introducing radio and telegraph communications, and radio waves and electrical pulses travel at the speed of light. Of course, as a practical matter, it takes time for an operator to convert a message into transmissible form, and, at the receiving end, for another operator to convert it back again. If the message has to be relayed, then effective transmission times are increased. But it’s still much faster than horse or ship.

Our climate is the result of the heating of the earth’s land masses, oceans and atmosphere by solar radiation, coupled with the rotation of the earth about an axis tilted relative to its orbital plane.

The amount of solar variation emitted by the sun varies, and it turns out that there’s a pretty good correlation between the number of sunspots and the solar output. All else being equal (and it rarely is), if solar output decreases, so will mean global temperature.

However, there is a more specific effect on radio communications. The solar radiation includes not only light photons, but also charged particles, and when those particles strike the earth’s atmosphere, they ionize some of the air molecules. When solar output is high, the degree of atmospheric ionization is higher, and it is easier to bounce radio signals off the “ionosphere” so that they can travel longer distances. The principle is explained in much more detail in Boatright, “Radio in the 1632 Universe,” Grantville Gazette 1.

Climate and Mining

Surface temperature doesn’t have much of a direct effect on underground mining; the temperature underground is mostly a function of latitude. However, it can affect how easy it is to get miners and their goods to the mine, and to ship off the ore. A good case in point is that in the nineteenth century, cryolite could be mined in Greenland for only a small part of the year.

Rainfall is another matter. Drainage was a serious problem in both European and Japanese mines, and I imagine that in periods of heavy rainfall, the problem was exacerbated.

Climate and Industry

Industrial production presupposes the existence of healthy indoor temperatures. It is already a common practice to heat homes and shops during the winters in colder regions of the world. Factories in those climes will also need heating systems, and, if it gets colder, they will require more fuel (most likely wood or coal).

Summers in warmer regions are more of a problem, because the only form of cooling is ventilation. True air conditioning requires up-time technology. Fortunately, in those areas affected by the LIA, summer is not a major concern.

The effect of increased rainfall is a more subtle one; more rainfall will be associated with more humidity, which means more problems with decay (wood) and rust (iron). This may increase industrial demand, but it also means that the maintenance costs will be higher.

Climate and Warfare

The conduct of war is also affected by climate, both indirectly and directly. If harvests are poor, it will be difficult to feed the troops and their work animals. If roads are muddy or snow-covered, troop movements will be slow. If the soldiers are not conditioned to the local climate, and properly dressed for it, there will be weather-associated deaths.

Climate begets weather, and one of the more piquant examples of the effect of weather on warfare was the January 23, 1795 capture of the Dutch fleet by the cavalry of the French Republic. It was trapped to the lee of Texel Island by ice.


The Up-Timers’ Perspective

The up-timers are coming from a West Virginia town. While Grantville is fictional, it is based on real-life Mannington, in north central West Virginia (Marion County). Climate data for Mannington goes back to 1948, but unfortunately it’s spread over three different weather stations. For nearby Fairmont, there’s continuous data from a single station.

Please note that interannual variability of even annual (let alone seasonal, monthly, or specific day of the year) temperatures is such that it is customary for weather services to calculate “climatological normals” over a thirty-year period.

Table 1 shows the sort of climate that the up-timers of Grantville are accustomed to. From this we can estimate seasonal average temperatures as follows: winter (DJF), 31.7oF (-0.2oC); spring (MAM), 51.0 (10.6); summer (JJA), 70.5 (21.4); autumn (SON), 54.2 (12.3). The average of the daily minimums for January was 20.4oF.


Table 1: Monthly Averages of Daily Means, Fairmont WV, 1971-2000










































Ditto, North Central WV (WV-02)















Standard Deviations of the Monthly Means, North Central WV (WV-02)















(Climatography #81, #85)

Fairmont (ZIP code 26554) was in the 1990 USDA Plant Hardiness Zone 6A (average absolute annual minimum temperature in range -10 to -5 o F, -20.6 o C to -23.3 o C), and in zone 6-7 of the 2006 Arbor Day Foundation update.

In this part of West Virginia, the first freezing temperatures (end of the growing season) is typically in the first half of October, and the last freezing temperature (preceding spring planting) in the first half of May.

A Global Overview of the LIA

In 2002, Mann presented a figure comparing temperatures for the period 1000-2000 for eight different parts of the world. Mann considers the LIA to be 1400-1900, and my comments are based on the reconstructed annual means. I will call an LIA “low” if the temperature was less than the lowest value for that region during 1000-1400.

Northern Hemisphere: the lows are in the late-16th, late-17th, and late-19th centuries, with highs in the early 17th and mid 18th centuries.

West North America: the deep lows are in the late-16th and mid-19th centuries, and a shallower but broader low appears in the 17th. The highs are in the early-15th, mid-16th and late-18th centuries.

Subtropical North Atlantic: there’s a broad shallow low centered on 1700, and a high in the early- and mid-16th century.

Western Greenland: The entire LIA was warmer than in the late-14th century, but at its warmest in the early-15th and coldest in the late-17th and late-19th centuries.

Central England: The LIA saw a long decline to the low of late-17th century, then an improvement in the early-18th century. Temperatures remained well below the broad peak of the 13th century.

Fennoscandia: the deep low is just after 1600, and temperatures gradually recovered to a broad peak in the late-18th and the whole 19th centuries.

Eastern China: the biggest temperature drop of 1000 – 2000 was before the LIA, in the 12th century. During the LIA, temperatures remained at or above their 14th century levels, with a broad peak in the 19th century.

Tropical Andes: the LIA really began around 1500 here, but there were no sharp lows. The lowest points are in the late-17th and late-18th centuries. The early-17th century was cooler than the 15th century but otherwise unremarkable.

Thus, the LIA was not simply a four-century cool period; it included warmer and cooler intervals, and these weren’t synchronous between regions. However, it has been contended with some justice that it was a period of greater temperature variability.

1630s Europe: Historical Accounts

We can learn a lot about past climates from historical records. At the very least, they speak directly to the real-life consequences of weather conditions (droughts, floods, freezes, heat waves, and storms). And in some cases the historical records provide quantifiable information (e.g., the dates that particular lakes or rivers froze or thawed, the dates of harvesting grapes or other crops) that can be correlated with overlapping instrumental records so that the older temperatures may be inferred.

While our interest is particularly in the 1630s, we will from time to time look back at dates that would have been in living memory, and forward to the 1640s.

Pfister has constructed, by rating the severity of temperature and rainfall extremes in documentary accounts and weighting them together, an index of climate impact on European agriculture. There were major peaks in 1569-73, the late 1590s, 1614, and 1626-29. 1628 was a “year without a summer.” (cp. Battaglia). “After 1630 the level of climatic stress drops substantially.” The next peak, in the 1640s, was of about half the magnitude of the one in 1626-29.

Temperature increased from the 1620s to the OTL 1630s, and the number of witchcraft trials in eleven regions of Europe, standardized relative to the regional means, declined. In the OTL 1640s, they increased again, to higher than the 1620s level (Oster, Fig. 1).

Great Britain. According to Wikipedia/River Thames Frost Fairs, in the 17th century, the Thames froze over at London in 1608, 1621, 1635, 1663, 1666, 1677, 1684 and 1695. With particular regard to the winter of 1635, the frost was severe from December 15 to February 11. It was followed by a warm and moist spring, and a very hot and dry summer and autumn. But the following winter (1635-36) was unseasonably warm. 1637 was also cold. The summers of 1636, 1637 and 1638 were all hot and dry (Marusek 116; LambCPFF 568).

During the LIA, the 25-year average of the English price of wheat increased from its low around 1500 to a high around 1650, then dropped to a shallower low in the late-18th century, and then climbed to a greater high in the early-19th century (Flohn 44; LambCPPF 462).

Note that during the coldest parts of the LIA (which for England was the late-17th century), the growing season was shortened by 1 – 2 months compared to that of modern England (Mandia).

Scandinavia and the Baltic. Historical climatologists have found records of the date of ice break-up at the harbor of Riga (Latvia) going back to 1529. We know that in the 1620s, 1620-21 was a severe winter, 1622-23 was average, and 1625-6 was mild. And in the 1640s, 1642-3 was severe, 1648-9 was average, and 1649-50 was mild. But the data for the 1630s are missing. The average break-up date is March 24 in a mild winter, April 3 in an average one, and April 12 in a severe one, but the variability is fairly high. (Jevrejeva). For the 17th century, the earliest date was Feb. 2, 1652 and the latest May 2, 1659 (LambCPFF 587).

There is also data, complete from 1600 on, for ice-breakup at Tallinn (Estonia), which is near the average western limit of the ice cover in the Gulf of Finland. (Tarand 192). The means for 1597-1629 were year-day 97.4 for Riga and 106.18 for Talinn, and for 1630-1662, they were 80.25 and 99.73 respectively. The estimated winter air temperatures for Tallinn were -5.84oC and -4.72oC for the two periods. And the “Ice Winter Severity Index” for the Western Baltic dropped from 0.73 to 0.44 (it was 0.02 in 1988-93). (Eriksson; Tarand 192).

Alas, the post-1622 Great Sea Toll records for Stockholm, recording the dates of first arrival and last departure for each shipping season, were requisitioned by the Swedish Army as—brace yourself—wadding for artillery. Nonetheless, useful records relative to the shipping industry have survived, and the climate observations from these records have been scaled and calibrated with overlapping instrumental data to reconstruct winter temperatures for Stockholm. These reveal that 1614-23 (-2.43oC) and 1624-33 (-2.20oC) were the second and fourth coldest decades since 1500. (The dangers of relying too much on the generic Little Ice Age label are shown by the fact that one of the five warmest decades, 1734-43, is within the conventional LIA.) The decade of 1634-43 was a bit warmer than 1624-33.

I have found reports of crop failures in Norway in 1632 and 1634 (GroveLIAAM, 67). These are probably attributable to the proximity of glaciers. The 1742 report of the court of inquiry on Elekrok stated “it was apparent to us that it was the nearness of the glacier which is the cause of crop failure on this farm . . . the ears on the side towards the glacier . . . were quite brown, and the other side green . . . .” (71).

The Netherlands. In East Friesland, on Sept. 1, 1637, there were great floods (Marusek 116). (The specific date won’t be repeated in the new time line, but there may still be a propensity to flooding that autumn.)

For wheat prices, the first LIA climb came later than for England, possibly in the 1640s. Otherwise, the fluctuations were similar to those for England but smaller.

Germany. The walls of historic buildings at Tonning on the west coast of Schleswig-Holstein reveal that the flood of Oct. 11, 1634 reached a height of four feet above the ground surface. (LambCHMW 17). On Norstrand Island, 6,123 people drowned, and 50,000 livestock were lost (Rabeljee). This flood is mentioned in Grantville literature, but the up-timers didn’t think to warn King Christian about it because they assumed it would be “butterflied away.” Instead, it came ahead of schedule. See Boyes, “A Great Drowning of Men,” Grantville Gazette 28.

The price of rye in Germany over four centuries has been analyzed. Peaks corresponded to a poor harvest—this could be because of climate, or because of warfare. Considering just 1590-1650, there were small peaks in 1590 and 1610, moderate ones in 1626, 1634 and 1649, and a large one in 1622. However, the worst one of all was that of 1816 (the “year without a summer”) (Mandia, Fig. 17). Other than in 1634, the 1630s appear to have offered cheap rye—albeit not as cheap as in the “good years” of the 16th century. The high prices in 1634 were probably attributable to plague (Pfister2007, Fig. 8).

France. On October 6, 1632, southern France was so cold that sixteen of Louis XIII’s bodyguards died from exposure (Marusek 115). The winter of 1638 was also severe; in Marseilles, the “water froze around the ships.”(116).

Wine grapes will reach maturity more quickly if the growing season (April-September) is warm, than if it is cool. Based on the extensive wine harvest data, the summers of 1634-39 were warmer than the 1599-1791 mean (Ladurier; Chuine).

French wheat prices followed a pattern similar to that of British, but the fluctuations were more moderate (Flohn 44).

Switzerland. Based on historical documentary evidence, Pfister constructed crude thermal (warm months-cold months) and wetness (wet months-dry months) indexes for Switzerland. The 1630s appear to be a little on the warm side, and markedly on the dry side. The 1640s were colder, although nowhere near as cold as the 1670s, 1690s, or 1810s, and not as dry (LambCHMW 204).

In the Alps, we have the very visible evidence of the advance and retreat of the glaciers. Unfortunately, in the 17th century, we do not have good maps of their positions, and hence we have to rely on diaries and legal documents. In 1600-19, there are repeated descriptions of the destruction of houses, the failure of crops and the decline in tithes as a result of glacial advance (Ladurie 143ff).

It appears that the glaciers were more quiescent in the 1620s and 1630s, but they remained dangerous. A third of the cultivable land at Chaimonix was destroyed in 1628-30, and the Mattmarksee (influenced by the Allalin glacier) flooded in 1620, 1626, 1629, and 1633 (Aug. 21). In 1636, the people in the valley of Randa thought that the whole Zermatt glacier was coming down on top of them; forty people were killed by ejecta. In the 1640s, there were new glacial advances. In May, 1642, the Les Bois glacier was reportedly moving by “over a musket shot every day.” Such ominous developments led to the famous June 1644 procession, led by the local bishop, to seek divine intervention to hold the glacier at bay (Ladurie, 165 – 173).

Wine harvest dates for fifteen locations in the Swiss Plateau and northwestern Switzerland have been used to reconstruct April-August temperatures. Harvest on year-day 285 indicated temperatures identical to those of the base period 1961-90; earlier harvests implying warmer temperatures. In 1632-33, temperatures were a little below base, whereas in 1634-39 they were higher, peaking at 1.78oC higher in 1638. The temperature anomalies in 1640-43 were negative (Meier).

Northern Italy. The second quarter of the seventeenth century was not marked, as was the first one, by any “great” winters (enough for large bodies of water to have ice thick enough to support people) or even “severe” winters (causing the death of animals and plants) (Alfani Graph 1.3). The flooding of tributaries (Tanaro and Bormida) of the Po was perhaps half as common as in the preceding quarter-century, but nonetheless more common than in the next one.

In 1629, a landslide, triggered by heavy rain, caused loss of life and property in the hamlet of Onera. In 1629-30, a plague epidemic killed about 27% of the population of Northern Italy, but the extent to which climatic factors contributed to its occurrence remains in dispute (Alfani). In 1632, there were complaints about both heat and drought (Marusek 115). In the lower Po valley, cereal yields were “seriously reduced in the period 1590-1630, especially.” That was, of course, attributable to the flooding (Grove 129).

The Italian price of wheat in the LIA reached a peak just after 1600, then descended to a broad low in the 18th century, then climbed more moderately to twin peaks in the 19th (Flohn 44).

Spain. The period 1575-1650 was “generally wet,” at least in the southeast. 1617 and 1626 were “deluge” years, and “catastrophic floods were unusually frequent between 1571 and 1630, especially in Catalonia.” (Grove 129). There was major flood activity in 1630 – 1650, too (Llasat Fig. 5).

The Black Art of Reconstructing Past Climates

Crude thermometers appeared in the 17th century, and our oldest continuously monthly temperature records date back to 1659 (for central England). For that region, the coldest winter was in 1684, the coldest summer in 1725, and the coldest year overall was 1704 (Manley). Other early records are those for Berlin from 1697, for Hoofddorp and Zwanenberg/De Bilt in the Netherlands from 1706 and 1735, respectively, for Uppsala (Sweden) from 1739, and for St. Petersburg from 1726 (Flohn).

Clearly, this direct data doesn’t tell us anything about what the temperatures were in 1631-39. However, it does help in calibrating “proxy” data.

A “proxy” is any observable variable of the fossil record (this term used in a broad sense) that can be reliably correlated with direct temperatures for part of the range of the record, so that the historical temperatures can be reconstructed for the rest of that range.

For our purposes, it isn’t sufficient that the proxy be highly correlated with the actual temperature, it also must be “high resolution.” For example, if we couldn’t determine the age of a proxy value more accurately than the nearest decade, or if the proxy value reflected the temperatures over the preceding decade, then the resolution it offers is just decadal. We want resolution down to the annual level.

Here are some of the sources of high-resolution proxy data:

Ice Cores—the upper portion of an ice core exhibits a layered structure with annual variation; the light bands are formed by freshly fallen, clean summer snow and the dark bands are formed by old, dust-contaminated winter snow. The thickness of the light band is indicative of how much snowfall there was. Air bubbles in the ice preserve “fossil” air, in which the level of greenhouse gases can be measured. Also, oxygen isotope ratios are influenced by ocean temperatures. Obviously, ice cores are only available from a few parts of the world; notably Greenland, Antartica, and a few glaciers.

Tree Rings—the light colored layer grows in the spring and the dark colored one in late summer. Narrow rings are indicative of poor growth conditions, such as drought or severe winter. Tree ring data is available only where trees grow.

Corals—we can see annual variations in skeletal density and geochemical parameters. The light layers are from the summer and the dark layers from the winter. Oxygen isotope ratios are indicative of ocean temperatures. The most useful corals grow in shallow tropical waters.

Lake Sediments—these may exhibit seasonable variations (varving) in runoff sediment composition, which in turn are the result of summer temperature, rainfall, and winter snowfall.

Boreholes—the variation of temperature with depth has a detectable relationship to the history of temperature at the surface.

Speleotherms—these are stalactites, stalagmites and flowstones. Some provide annual resolution, as a result of visually detectable lamination, or a seasonal variation in trace elements. Layer thickness is related to surface rainfall and cave air temperature.

Historical accounts—these are most useful if they provide some kind of quantitative information.

There are technical problems with working with proxy data, but consideration of those problems is outside the scope of this article.

Climate Reconstructions: The North Atlantic Oscillation

In the mid-latitudes of the North Atlantic, the prevailing winds are from the west. These were convenient for mariners returning from the New World. However, those winds are also important because they bring moist air to Europe.

The direction and strength of the prevailing winds are controlled by the position and strength of a persistent low-pressure system over Iceland (the Icelandic Low), and a persistent high-pressure system over the Azores (the Azore High).

The atmosphere alternates between a state in which the pressure difference widens (positive phase, NAO+) and one in which it narrows (negative phase, NAO-). There are a number of ways the NAO may be quantified, but the simplest is as the normalized difference in pressure between a station in the Azores (or in Portugal) and one in Iceland. There is no significant periodicity in the switching between NAO+ and NAO-.

In NAO+, the westerlies are stronger, and more and stronger winter storms cross the north Atlantic, on a more northerly track. Temperatures are above average in the eastern United States and in northern Europe, and below average in northern Canada and Greenland and often in southern Europe, northern Africa and the Middle East. There is also above average precipitation in northern Europe, and below average in southern Europe. In NAO-, the effects are reversed.

The effects are strongest in winter. (NWS-CPC).

The North Atlantic Oscillation index has been reconstructed, on a seasonal basis, for 1500 – 1658 and monthly for 1659 – 2001 (LuterbacherNAO). Table 2-1 shows its behavior for 1630 – 39. It can be seen that it was mostly in negative phase in that decade.


Table 2-1: Reconstructed North Atlantic Oscillation Index

winter spring summer autumn




















































Climate Reconstructions: European Annual Average Temperatures

Looking first at reconstructed mean annual temperatures for Europe generally, Table 2-2A shows how the 1630s (with the years 1999-2000 for comparison) shape up.


Table 2-2A: European Annual Mean Temperatures
Year Annual Mean Temp oC Coldness Rank, 1500-2004 (505 years)
1630 8.44 362
1631 8.352 325
1632 7.872 105
1633 8.268 279
1634 8.05 164
1635 7.498 30
1636 8.371 330
1637 8.05 165
1638 8.317 305
1639 8.295 294
1999 9.436 500
2000 9.664 497


If we consider just the temperature column, it’s clear why the up-timers feel a chill in the air. However, the coldness ranks (30-362) provide some perspective. And it’s worth comparing those temperature to the averages (Table 2-2B) for each century, for the Maunder Minimum (1645-1715), the whole LIA (1500-1850), and the modern period (1851-2004).


Table 2-2B: Multiyear averages of European Annual Mean Temps
Years Average Annual Mean Temp o C

















We can see that only three years were below the average (8.2oC) for the LIA. The worst year of all (1635) was the 30th coldest year for the period studied (1500-2004). It was not the coldest year in living memory; that was probably 1573 (7.0oC, 2nd coldest), and down-timers will also remember 1587 (14th), 1595 (10th), 1600 (5th), 1601 (8th), 1608 (6th), and perhaps also 1565 (13th) and 1569 (9th).

So, yes, we are in the LIA—but not in the worst decade, by any means.

Climate Reconstructions: European Seasonal Average Temperatures

There are several references to the severity of winter in 1632 universe canon. Watching TV in 1633, Joyce and Gary find that “the news was about broken armies, new business, and, of course, the weather and what the little ice age meant to their future.” (Huff and Goodlett, “Wish Book” Grantville Gazette 12). In January 1634, Eric Krentz tells Thorsten Engler, “I always hated January even before an up-timer told me we’re in the middle of what they call ‘the Little Ice Age.’ ” (Flint, 1634: The Baltic War, Chapter 14). Later, in late March 1634, Admiral Simpson is “a bit surprised that the river [Elbe] hadn’t frozen, although intellectually he’d understood that the past winter hadn’t really been as cold as it had sometimes felt, Little Ice Age or not.” (Chapter 31). In Flint and DeMarce, 1635: The Dreeson Incident, we are told that “Winter in Thuringia during the Little Ice Age encouraged the layered look.” (Chap. 41).

So, how did LIA winters (and other seasons), compare to those the up-timers would have been acclimated to, and just how bad were the 1630s as compared to other parts of the LIA?

We define the seasons as DJF (winter, December-January-February, and I assume that winter for 1630 starts with Dec. 1629), MAM (spring), JJA (summer) and SON (fall). Table 2-3 provides the reconstructions for each of 1630-1639, the actual values for 1999 and 2000, and averages of values for the decade 1630-39, the 30-year period 1620-1649, 1645-1715 (the Maunder sunspot minimum), 1500-1850 (LIA) and 1851-2004 (Post-LIA). The coldest (bold blue) and warmest (italic red) winter, spring, summer, fall and entire year are marked.


Table 2-3: Europe, Seasonal Mean Temperatures (o C)



-0.033 6.934 17.867 9.013 8.440


-1.214 7.320 18.044 9.280 8.352


-1.134 6.863 17.320 8.458 7.872


-0.080 6.971 17.351 8.851 8.268


-1.187 7.097 17.684 8.625 8.050


-2.932 6.396 17.532 9.017 7.498


-0.877 7.958 17.743 8.681 8.371


-1.784 7.416 18.070 8.518 8.050


-0.998 7.763 17.945 8.579 8.317


-0.433 7.043 17.230 9.360 8.295


-1.067 7.176 17.679 8.838 8.151


-0.195 8.513 18.829 10.121 9.436


0.743 8.947 18.177 10.475 9.664


0.333 8.083 18.143 9.283 8.955


-1.312 6.784 17.576 8.973 8.118


-1.125 7.045 17.651 9.03 8.176


-0.845 7.297 17.617 9.096 8.292


I have calculated, but not shown, the seasonal coldness ranks. Looking first at winter (DJF), 1635 was the 32nd coldest in the study period. The next worst was 1637, ranking 129th. The mildest winter of the decade was that of 1633, ranking 408th. The LIA mean was -1.125C, so five years were better and five were worse, and the mean for our decade was a bit milder.

The springs (MAM) ranged in rank from 62nd to 459th coldest, the summers (JJA) from 92nd to 424th, and the falls (SON) from 60th to 372nd. For all three seasons, the mean for our decade was higher than the mean for the LIA.

Surprisingly, the mean for summer 1630-1639 was even higher than the mean for the post-LIA (1851-2004) period, although of course still cooler than the summers of 1999 and 2000.

We are lucky to have missed 1628, which was the thirteenth coldest summer (16.8oC) of 1500-2004.

Climate Reconstructions: Central European Monthly Average Temperatures

Monthly temperatures have been reconstructed for central Europe in 1630s, based on documentary evidence (mostly from sites in the present Germany, Czech Republic, and Switzerland). Unfortunately, these are available just as anomalies, that is, the difference between the actual temperature and the average(s) for the base period 1961-1990 (Dobrovolny). I emailed Dr. Dobrovolny, asking him for the base temperatures, but he didn’t reply. So I used the KNMI Climate Explorer to download the CPC GHCN/CAMS t2m analysis (land) data, gridded at 0.5° intervals, and calculated the 1961-1990 base for central Europe myself. I assumed that Dobrovolny defined central Europe the same as his coworkers did in LuterbacherSLP.

In Table 2-4, I show first my derived monthly temperatures for each year of 1630-39, and the average and standard deviation for the decade. Next, I provide my 1961-99 base numbers. There’re no guarantees that Dobrovolny had exactly the same base averages; his region and his modern data could have differed from mine. But since I stated the bases I used, you can reconstruct Dobrovolny’s anomalies by subtracting them out. Finally, I present the mean and standard deviation for the period 1766-1850, from Luterbacher’s monthly gridded reconstructions.

Table 2-4: Central European Monthly Mean Temperatures (Anomaly + Base) oC















































































































































Base Averages for LuterbacherSLP Central Europe, 1961-99, oC














Please note that with the exception of 1630 and the first four months of 1631, this monthly data is subject to perturbation by the RoF. Since it’s time-averaged data, the impact won’t be as severe as for daily weather, but there will be some effect.

Climate Reconstructions: Mapping Post-RoF European Seasonal Average Temperatures

Of course, pan-European averages are all well and good, but different parts of Europe would no doubt fare differently. Fortunately, Luterbacher’s climate reconstruction provides reconstructed seasonal temperature for each point on a 0.5 degree by 0.5 degree grid, over the range 25W-40E longitude; and 35N-70N latitude.

It’s said that a picture is worth a thousand words, and I have created some revealing images from Luterbacher’s gridded temperature data. Using Climate Explorer, I have compared the decade 1630-39 with 1990-99. The figures compare the decades for (Fig. 1A) the entire year, (1B) winter, (1C) spring, (1D) summer and (1E) autumn.

And using the National Climatic Data Center’s visualization tool, I have created comparisons of the coldest and warmest winters (2A), springs (2B), summers (2C) and autumns (2D) of the 1630s. Note that each season has a different temperature scale:

Winter (DJF): coldest 1635, warmest 1632, scale -20 to +10C;

Spring (MAM): coldest 1635, warmest 1636, scale -5 to +25C;

Summer (JJA): coldest 1630, warmest 1637, scale +5 to +35C;

Fall (SON): coldest 1635, warmest 1630, scale -5 to +25C.

Which parts of Europe are unusually hot and which are unusually cold is very strongly influenced by the position, areal extent and persistence of the high and low pressure areas (see North Atlantic Oscillation) and the position and strength of the jet stream. Stagnant (blocking) patterns lead to persistent weather conditions that influence monthly and even seasonal averages. On the west side of a stationary NH high, warm air is pushed north, and on the east side, cold air is dragged south. So you may be warmed or cooled depending on where you stand. Moreover, a slight shift in the location of the blocking pattern from one year to the next might mean that you face extreme cold in the first year and extreme heat in the second (LambWCHA 110).

Climate Reconstructions: Post-RoF Grantville Seasonal Average Temperatures

The place of greatest interest to the up-timers is, of course, the location in Thuringia where the RoF deposited Grantville. The center of the RoF was at approximately 11o16′ east longitude, 50o40’12” north latitude. The closest Luterbacher grid point is 50.75N, 11.25E, and the reconstructed seasonal temperatures for this location are in Table 2-5A (with comparison to pre-RoF Grantville at the bottom of the table).


Table 2-5A: RoF Site in Thuringia,

Seasonal Mean Temperatures (oC)



0.600 6.110 16.780 7.930 7.855


-1.800 7.300 17.330 8.390 7.805


-0.810 6.360 15.300 6.730 6.895


0.700 6.730 15.340 7.370 7.535


-1.310 6.680 16.280 7.420 7.268


-4.220 5.530 16.010 8.060 6.345


-0.630 8.520 16.600 7.380 7.968


-2.480 7.980 17.180 7.020 7.425


-1.070 8.550 16.900 6.990 7.843


0.500 6.490 15.170 8.470 7.408


-1.387 6.622 16.052 7.692 7.245


-1.152 7.025 16.289 7.576 7.435


-1.673 6.536 15.997 7.705 7.141


-0.705 7.106 15.916 8.303 7.655


-0.062 7.542 16.214 8.178 7.968


0.517 8.262 16.734 8.221 8.433

pre-RoF Grantville, ditto


-0.2 10.6 21.4 12.3 10.9


It can be seen that in OTL 1620-49, the growing season (April-September) was probably about 4-5oC. colder than in 1971-2000 Grantville. If extremes moved downward the same amount, that probably wouldn’t shift Grantville into a new plant hardiness zone (that would require an 18oC change.)

Climate Reconstructions: Magdeburg Seasonal Average Temperatures, 163039

Magdeburg is at 52o07’N, 11o38’E, and the closest grid point is 52.25N, 11.25E. The data for that grid point are in Table 2-5B.


Table 2-5B: Magdeburg, Seasonal Mean Temperatures (oC)



1.980 6.980 17.590 9.120 8.920


-0.340 8.090 18.210 9.510 8.870


0.510 7.190 16.090 7.990 7.950


2.100 7.540 16.090 8.530 8.570


0.130 7.510 17.090 8.730 8.370


-3.130 6.400 16.930 9.270 7.370


0.800 9.250 17.550 8.650 9.060


-0.930 8.710 17.890 8.300 8.490


0.380 9.260 17.650 8.240 8.880


0.800 7.340 16.050 9.560 8.440


-0.081 7.452 16.852 8.869 8.273


0.23 7.83 17.11 8.79 8.49


2.17 9.83 17.9 10.77 10.13


3.27 10.53 17.13 10.9 10.46


1.98 9.27 17.63 9.42 9.58


It is possible to extract the reconstructed temperatures for other locations in Europe, too, given their latitude and longitude. The necessary data set and format information are here: Please note I had to write a program to extract the data, because Excel can’t import 18,000 columns of data . . . .

Climate Reconstruction: Plausible Grantville Monthly Average Temperatures

Unfortunately, I don’t have gridded monthly temperature reconstructions covering the 1630s. Mark Twain once said, “there are three kinds of liars: liars, damn liars, and statisticians.” We can make an educated guess as to what the monthly temperatures were, using statistics for other time periods. There are a number of ways that this can be done. I assumed that the relationship of monthly to seasonal temperatures for Grantville was the same as for Central Europe.

Or, in mathematical terms,

Grantville average for that month= Grantville average for that season (from LuterbacherTemp) + adjustment, where the adjustment was Dobrovolny’s central Europe average for that month – central Europe average for that season.

Table 2-6A provides my reconstructed monthly temperatures for the location that Grantville was transported to. For convenience, I also repeat the climatological normals for Fairmont, West Virginia.

Table 2-6A: Plausible Grantville (Thuringia) Monthly Average Temperatures (oC)




























































































































































sd oC


























sd oF


























sd C


























sd F













Monthly Average Temperatures for ‘Grantville” West Virginia, 1971-2000















29.20 32.00 41.20 51.10 60.60 68.30 72.20 70.90 64.40 53.10 43.10 34.00


Bear in mind that these monthly numbers, even if accurately reconstructed, are the likeliest climate statistics to be corrupted by the “butterfly effect” of the RoF. So that’s another good reason to view them as general indications rather than gospel truth.

What I thought most noteworthy about them was how fast temperatures dropped off during autumn and rose during spring.

Table 2-6B provides the average and standard deviation, over 1766-1850, of Luterbacher’s reconstruction of monthly mean temperature at the same location. These, of course, reflect a different time period, but they save us the trouble of trying to convert temperature anomalies into absolute temperatures.


Table 2-6B 1766-1850 Average of Luterbacher Mean Average Temperature,

Grantville in Thuringia (50.75N, 11.25E), oC







































Grantville Growing Degree Days

We can estimate the number of growing degree days for a given location and year, for whatever base is appropriate to a particular crop. (Schenkler; Thom). Using the estimated monthly means and standard deviations from Table 2-6A, then by Thom’s method we get the dramatic results shown in Table 2-7.

Grantville Extreme Temperatures

We also can make some educated guesses as to typical daily minimum (usually nighttime) and maximum (usually daytime) temperatures. The climatological norm of the monthly mean of the daily temperature range (maximum-minimum, DTR) varies from month to month, and is affected by latitude (which determines solar radiation variation), distance from the shore (affecting exposure to sea breezes and degree of low-level cloudiness), and precipitation. In the northern hemisphere, the DTR peaks at 20-40°N. In modern Europe, at latitude 50°N, the DTR is about 12°C in July and 7°C in January (Geerts Fig. 3). Moving inland more than 100 km increases the DTR by about 2°C. Those rules should apply to Grantville in Thuringia.

Unfortunately, there’s a catch: the DTR can change as the climate changes. In particular, “the blocking action of greenhouse gases would be most effective where outward radiation was most important for cooling the Earth: warming would come especially at night” (Weart). So the DTR of the late-twentieth century is probably smaller than that of the seventeenth, when greenhouse gases were at lower concentrations and thus had less of an upward influence on nighttime temperatures.


The USDA plant hardiness zones are defined on the basis of the climatological norm of the annual minimum—the lowest daily minimum recorded during the course of the year. Unfortunately, to calculate that, by Monto Carlo methods, we need to know not only the monthly mean and standard deviation for (at least) January, but also the correlation of one day with the next.

Plant Hardiness Zone Maps have been created for modern Europe that use the same scale as the USDA maps (Heinze), and based on 1881-1930 data (although Lorek says that including 1931-1960 data would have an insignificant effect); Grantville post-RoF is in zone 6b (average January minimum of 20.5 – 17.8oC, -5 to -10oF) and Magdeburg in 7b (14.9 – 12.3oC, 10 to 5oF).


A change in climate may simply shift the mean temperature, and leave the variability intact. Or it can alter the variability, too. If for example, it was not only colder in the 1630s than in the 1990s, the variability increased, then the likelihood of frosts would be much greater than you would expect by just considering the mean. To complicate matters further, there’s no guarantee that the “cold” and “warm” tails of the distribution will be affected by the same amount or even in the same direction.

Scientists have looked at both twentieth-century observational data (Kurbis, Moberg, Easterling, Karl, Michaels) and runs of global climate models, and about all I can conclude from this is that it’s not safe to assume that the variability of temperature was the same in the early-17th century as it was in the modern period. Unfortunately, I have no easy way to predict how it was different and therefore I must just admit that this is something the authors may easily play around with.

Climate Reconstructions: European Seasonal Precipitation, 1630-39

Temperature, of course, is only one of the two principal dimensions of climate; the other is precipitation. Pauling reconstructed precipitation across Europe in 1500-1900, and his 1630-39 results are in Table 2-8. Fig. 3 compares the average for 1630-39 to that for 1990-99.

Table 2-8: Europe: Reconstructed Seasonal Precipitation (mm)

Europe WI SP SU AU Annual


167.56 130.01 159.74 178.18 635.49


155.68 128.95 158.8 164.46 607.89


164.53 130.54 172.06 176.05 643.18


154.35 131.16 172.33 176.2 634.04


164.54 126.03 167.76 159.54 617.87


159.54 140.85 166.61 174.4 641.4


178.24 120.74 170.53 188.84 658.35


167.17 135.74 163.61 186.14 652.66


165.2 135.54 165.76 184.7 651.2


158.95 133.94 176.8 163.39 633.08


163.58 131.35 167.4 175.19 637.52


6.94 5.61 5.72 10.04 15.63


I provided the standard deviation, as well as the average, for 1630-39, and it is apparent that the annual variation from year to year for a given season was fairly small. However, because of the North Atlantic Oscillation, it’s not unusual for northern Europe to be dry when southern Europe is wet, and vice versa.

Climate Reconstructions: European Seasonal Sea Level Pressure Patterns, 1630-39

Reconstructions are available of the average sea level and upper air pressure (in millibars) for each season in 1500-1658, and thereafter monthly (Luterbacher2002). For reasons I will explore in part III, this pressure history is more likely to get buffeted by the RoF than the temperature data, so don’t put a lot of faith in it. Still, the patterns shown will remain plausible patterns. For 1630-4 and 1636-39, we had a winter high near the Azores and low near Iceland, with these highs and lows weaker in the other seasons. The Icelandic Low was distinctly weaker and displaced in the winter of 1635.

Radio Communications in the 1630s

Sunspot number normally varies according to a somewhat irregular 11-year cycle. However, there have been several periods of prolonged depression of sunspot number, notably the Sporer Minimum (1460-1550), the Maunder Minimum (1645-1715), and the Dalton Minimum (1790-1830), all of which correlate with cold temperatures. In the old time line, of course, no one had to worry about radio communications during these minima!

While it may sound as though we don’t have to worry about the Maunder Minimum yet, that’s not so. You see, the sunspot number is already on the decline. The total number of sunspots in the period 1630-1640 (eleven years) was 185.4. In contrast, in 1980-1990, the total was 956.1. The period 1645-1715 was simply worse, with no sunspots at all in most years.

So the decade of the 1630s may be considered the “slippery slope” down to the Maunder Minimum. And that means that (with the exception of 1638-9 and 1642) we will be facing progressively greater limitations on the range of radio communications.

The Arctic

The great circle route is the shortest distance between two points, but it can require you to sail at dangerously high latitudes. While the cold can cause frostbite and sap strength, the greater danger is from sea ice.

In the North Atlantic, the principal pathways by which ice can descend from the Arctic Circle are the Davis Strait and Baffin Bay between Newfoundland and Baffin Island to the west and Greenland to the east, the Denmark Strait (Greenland Sound) between Greenland and Iceland, and the Norwegian Sea between Iceland and Norway. Nowadays, Arctic sea ice reaches its maximum overall extent in March (Polyak), and I suspect this was likewise true in the 17th century. However, note sea ice can expand in the Greenland sound while contracting in the Davis Strait, and vice versa. In a “normal” severe year, the ice could surround Spitsbergen and bisect Iceland. It only rarely reaches the south coast of Iceland. (Ogilvie).

Looking at the documentary evidence-based sea-ice index for Iceland, filtered through a 15-year low-pass filter, the 1630s and 1640s exhibited values under 2, while the filtered index climbed above 5 in the 1780s, 1810s, and 1830s. (Ogilvie). So the 1630s are not especially bad insofar as sea ice is concerned, although in one of the years (1632? 1633?) it jumped to level 6. An earlier (Koch 1945) study says that there was drift ice at the coast for 24 weeks in 1633, 26 in 1638 and 17 in 1639 (LambCPFF 583). Consistently, GroveLIAAM (22) states that “the 1630s . . . which were cold on land, saw little sea ice . . . .”

For the waters around Greenland, based on GISP2 ice core chemistry, the sea ice levels increased more or less steadily throughout the first half of the 17th century. However, the levels in the early- and mid-19th century were higher (Dugmore Fig. 7).

Jabe McDougal’s musings continued, “After the high school had been saved from the Croat raiders, there had been a wave of interest in Swedish and Scandinavian history. Jabe had learned about the Little Ice Age and its presumed role in the death of the Viking colonies in Greenland.” (This role is now considered debatable, see e.g. Mann.)

While we aren’t interested in colonizing Greenland, it does have resources of interest. In year 1633 of the new time line, the Dutch metal and armament magnate Louis de Geer sent an expedition to Ivigtut (61o12’N/48o10’W) in southwest Greenland, at the mouth of Arsuk Fjord, to search for cryolite (the flux needed for efficient electrolytic production of aluminum metal from aluminum oxide) (Mackey, Kim, “Land of Ice and Sun,” Grantville Gazette 11). The most obvious objection to an expedition of this type would have been that the local climate, oppressively cold even today, would have been far worse during the Little Ice Age.

Well, maybe. But as Kim pointed out when the story was in slush, while it was certainly cold, there was evidence that in the 1630s, it was no colder than when cryolite mining began (1854). Production was 14,000 tons in 1857-67; 70,000 in 1867-77 (Johnson’s).

While I don’t have air temperatures for Ivigtut per se, ice core data for Site J (66o51.9’N/46o15.9’W, 2030m) in South Greenland, inland, was used to reconstruct June temperatures for Jakobshavn (69o13’N,51o6’W,30m) on the west coast. This was -0.11oC in 1633, 0.08 in 1634, 0.30 in 1635, and remained above 0oC for the rest of the 1630s. It was -0.17 in 1854 and was under 0oC from 1857-1875 (Kameda). Ivigtut should have been warmer than Jakobshavn.

In the 1632 universe, we also have a Danish colony on Hudson Bay founded in 1634. We can glean a bit of climate information from the reports of the expeditions that visited Hudson Bay or its southward extension James Bay. Hudson entered Hudson Bay in early July and was frozen into James Bay on November 10, 1610. Luke Fox entered Hudson Bay in late May, 1631. Thomas James entered Hudson Bay on July 16 and James Bay in early September, 1632.

Otherwise, unfortunately, we must decide which part of the post-1700 (Guiot) or 1750 (Catchpole; Ball) data is most analogous to the 1630s and 1640s.


The wetness/dryness index (norm 10) for Russia (35oE) in the high summer (July – August) is 11.5 for the 1630s and 7.5 for the 1640s. LambCPPF 562). The winter mildness/severity index ( norm zero) was -10 for the 1630s and -36 for the 1640s (the worst value over 1100-1969). (564). Brooks (249) says “In the 1640s . . . severe cold was reported for every month of winter, making this the coldest decade in Russian history since the twelfth century.”

North America

For an overview, a good place to start is the North American Drought Atlas, which includes Palmer Drought Severity Index (PDSI) values, for summer 1634-1639, based on tree-ring data (Cook). You can see that the northeast suffered a drought in 1634, which deepened in 1635. California was wet in 1635-6, but it and indeed the entire Pacific Northwest dried up in 1637-39. Mexico was generally quite wet. Unfortunately, rainfall patterns are likely to be perturbed by the RoF. Hence, take Fig. 4, which shows the pattern for 1635-37, with a very large grain of salt.

Figure 4

There is also a tree-ring reconstruction of annual and seasonal temperature and precipitation anomalies for the USA from 1602 on. From this we can see, for example, that in the northeast, both winter and summer 1634 and 1638 were relatively cool, while 1635-37 were characterized by relatively warm summers and relatively cold winters (Fritts).

The 1630s were probably not as bad as 1608; Champlain found bearing ice on the edges of Lake Superior in June 1608 (LambCHMW 230).

Virginia. We don’t know for sure why the Lost Colony of Roanoke Island disappeared, but it probably was at least partially attributable to 1587-1589 being the driest three years in eight centuries. The second, successful English attempt to colonize Virginia teetered on the edge of failure. The English colony at Jamestown was founded in 1607, and 1606-1612 was the driest seven years in a 770 year period (reaching a nadir of Palmer Hydrological Drought Index [PHDI] -2.323 in 1610). Despite its coastal position, Tidewater Virginia exhibited droughts in the future, too; there was a short one, for example, in the late 1630s (PHDI of -1.687 in 1637 and -2.67 in 1638) (Stahle).

Spring water temperatures have been reconstructed for Chesapeake Bay based on crustacea mineral ratios. Temperatures were high in the first quarter of the 17th century (higher, in fact, than in the Medieval Warm Period), but then declined, reaching a low in the mid-18th century. In our period, we have 16.6 (1638), 8.97 (1642) and 10.74 oC (1646). The most recent value was 12.5 oC (1995) (Cronin)

New England. In Eastern Massachusetts, the 1630s were cool and wet, while the 1640s were cool and damp. In southern New England, there were two dry growing seasons in each decade, but no years with floods. Note that the same year can be both; the 1780s had six dry years but eight with floods. (Baron).

Eastern Canada. Southwestern Ontario has been identified as having a cool dry climate from 1600 to 1750 (Buhay). At Quebec City, there has been a study of the ice bridge formation (IBF) rate on the Saint Lawrence River. The river at that point is one kilometer wide. Surprisingly, the IBF frequency in 1620-1800 was 16%, less than the 48% seen in 1801-1910. There were seven IBFs in 1620-1660, and none in 1661-1740. Contrast that with the 80% in 1866-1885. Thus, in our period of interest, southern Quebec is relatively warm! (Houle).

Tropical America

In tropical America, the seasons are wet or dry, not hot or cold. The seasonal variation in rainfall is correlated with the seasonal movement of the Inter-Tropical Convergence Zone (ITCZ), northward in the NH summer and southward in the NH winter. In NH summer, it’s over northern Venezuela, and that’s the rainy season.

In northwest Yucatan and in Venezuela, the general period 1500-1800 was marked by drier conditions, usually attributed to a “southward displacement of the ITCZ and hence reduced trade wind moisture supply to the Caribbean during the summer.” However, at Lago Verde at the Isthmus of Mexico, a “particularly wet area,” this change was apparently not enough to create a moisture deficiency. Indeed, 1600-1650 seems to have been a very good half-century for the tropical forest growth there, suggesting that the dry season was shorter than usual (Lozano-Garcia; Peterson).

On the other hand, in northeast Peru, the LIA was about 10-20% wetter than the 20th century, with the late 16th and the 17th centuries being the wettest period (Reuter).

In the Venezuelan Andes, there were glacial advances in 1180-1350, 1450-1590, 1640-1730, and 1800-1820. (Polissar). However, in the Cariaco Basin (offshore Venezuela), sea surface temperatures were higher in the 17th century than in the 20th (Reuter).

In central Chile, the 1630s were a bit drier than normal (LambCPFF 638), which would have made rivers easier to ford, but also reduced crop yield.


Information on African conditions is pretty limited. Grove (38) says that southern Africa experienced “sudden warming from 1500 to 1675.” In the Makapansgat Valley of South Africa, per speleotherm data, temperatures seem to have been slightly depressed (relative to 1961-1990 base) during the period ~1320-~1750 (Tyson fig. 3). As best I can tell, in the 1630s we were edging up from a low (about 0.5 o C below the base) reached around 1600.

In equatorial Africa, the 17th century was marked by droughts in the west and center, and (at least after 1625) wet conditions from Lake Victoria east (Russell; Verschuren). It has been suggested that megadroughts in West Africa were associated with anomalously strong Atlantic trade winds (Overpeck), which would have facilitated colonization of the Americas. The area of Timbuktu was punished by famines due to drought (in 1617-1743) and by great floods (in 1640-1672 and 1703 – 1738), often in the same year (LambCHMW 226).


Monsoons are seasonal changes in prevailing wind direction. The Monsoon is essentially the Mother of All Sea Breezes (and Land Breezes). Water warms and cools more slowly than land. It is common for coasts to experience a sea breeze (wind blowing from the sea toward the land) during the morning, the air over the land warming first. Then, in the evening, there is a land breeze in the reverse direction, the air over the sea cooling last.

In a monsoon, this happens on a giant scale, and the “breeze” persists for several months. The summer monsoon, being a giant sea breeze, results in high humidity. And the winter monsoon, being a giant land breeze, brings dry conditions. (Unless the winds sweep over intervening water, such as the Bay of Bengal, before they reach you.)

In the Indian Ocean, there is a summer monsoon, with winds from the southwest, and a winter monsoon, with winds from the northeast. The exact dates of arrival and departure of the monsoons varies depending on exactly where you are located.

There is absolutely no doubt that there were monsoons in the Indian Ocean and the western Pacific Ocean before the Ring of Fire; in fact, trade and agriculture depended upon them.

The failure of the monsoon can result in a megadrought, with substantial loss of life.

A network of tree ring chronologies has been used to reconstruction monsoon conditions for Sino-Indian Asia; Figures 5A-5D show the “Palmer Drought Severity Index” (red dry, blue wet) for the region for the years 1630-1641 (NOAA/MADAgrid). Don’t fret about the particular years, since the RoF will scramble that, but note the degree of variation from year to year.

Figures 5A -5D

El Nino/Southern Oscillation (ENSO)

In the Pacific, there is a alternation between two climate patterns, “El Ni-o” and “La Ni-a” over a period of three to seven years. These opposite states may persist for just a few months or for as much as two years.

In an “El Ni-o” state, there is (by definition) a warming of at least 0.5oC of the surface temperature in the east-central tropical Pacific Ocean. This weakens the trade winds of the South Pacific, and causes drought in the western Pacific and rainfall in the eastern Pacific. Off the coast of Peru, the upwelling of cold water is inhibited, and this results in fish kills. Winters in western South America tend to be much warmer and wetter than usual. The same is true, to a lesser degree, on the Pacific Coast of Central America, Mexico, and southern and central California. However, in the northwestern United States and western Canada, winters are warmer and drier.

We know from documentary evidence (mostly from Peru) that there was a very strong event in 1578 (comparable in strength to the 1982-3 event), and since then, strong ones in 1607, 1614, 1618, 1619, and 1624 (comparable to the 1972-3 event). If climate follows the old time line path, there will be strong ones in 1634-35 and 1640-41, and a moderately strong one in 1647 (Arteaga; Quinn).

Interdecadal Pacific Oscillation (IPO)

There’s also the IPO, which flips every 15-30 years. In a positive IPO, India experiences a weak monsoon and associated drought, and the American West Coast is wet.


A chronology prepared by Sindh historian M.H. Panwhar has some interesting climate-related entries for our period of interest:

“Famine in Gujarat and Deccan due to failure of monsoons. This famine was due to failure or rains in 1630 and excessive rain in 1631. People sold their children so that they may live . . . . Hides of catle [sic] and flesh of dogs were eaten, cremated bones of dead were sold with flour and cannibalism became common . . . . Three million people died between 1630-1633 in Gujarat.

“1636-1637. Punjab had famine . . . .

“1640. Heavy rain caused floods and destroyed crops in the Punjab and Kashmir, causing famine . . . .

“1640-44. Rains failed continuously in many parts of Northern India and famines occurred in Agra province . . . .

“1642. Famine occurred due to heavy rain and floods in the Punjab . . . .

“1646. Drought in Agra and Ahmedabad . . . .

“1647. Rains failed in Marwar. Famines, high mortality . . . .

“1648. Failure of rains in Agra area . . . .”

Southeast Asia

Newson (35) writes, “there is some evidence, notably from tree-ring data from Java, that during the seventeenth century some countries in Southeast Asia experienced unstable climatic conditions, perhaps linked to the ‘Little Ice Age’ in Europe, that included frequent dry periods that resulted in food shortages and famines. However, Peter Boomgaard suggests that climatic conditions were probably not so anomalous . . . ”


China experienced severe cold in 1629-43, as well as severe drought in 1637-43 (Brooks 269). Not coincidentally, the Ming dynasty fell in 1644. Climate extremes led to famine (the first big one was in 1630), which led to peasant revolts.

Famine also meant greater vulnerability to disease, and migrations to flee affected areas aided disease transmission. There were significant epidemics in the northwest beginning in 1633, and in the Yangtze valley in 1639. (Brooks 250ff).

In the Yangtze Delta, historical records show that 1635-1644 was marred by four flood events (years?) and six drought events. This was part of a larger trend; there were many flood and drought events in the period 1540-1670. In contrast, in 1495-1504, there was one flood event and no droughts (Jiang). Qiang says in 1550-1850, calamities “occurred by turns and sometimes, both drought and flooding occurred in the same year.” There was snowfall and frost on low ground in south China in 1635 and 1636, and the River Huai froze over in 1640 (LambCPFF 612).

Perhaps the most unique aspect of the Little Ice Age in China was the increase in reports of dragon sightings (Brooks 6ff); for example, “two dragons were spotted in autumn of 1643 . . . .” (14). Dragons were associated with water, and thus with storms and, more generally, bad weather. But they were both symbols of the emperor and celestial messengers. If people were seeing dragons, then what they were really perceiving was climatic evidence that the emperor had lost the Mandate of Heaven. And of course these reports in turn made it more likely that the emperor would lose support.


In Japan, 17th-century (and earlier) temperatures have been reconstructed on the basis of

—the dates of cherry blossom viewing parties in Kyoto

Overall, the average full-flowering date was day 105; the average for the 17th century was 106 (April 16), and for the 20-21c, 101 (April 11). The estimated March mean temperature for the 1630s was around 7oC, and the LIA low was around 6oC in late-17th century and early 18th century. A deeper low of ~5oC was inferred for the early-14th century)(Aono). In 1633, the cherry trees blossomed on April 8 (LambCPFF 607).

—the dates of freezing (December-January), buckling (Omiwatari, supposedly the footprints of a kami), and thawing of Lake Suwa in Nagano

Overall, the average freezing date was January 15, and the mean for the 1630s was about 12 days early, and that was actually one of the coldest decades recorded)(Lamb 256). By way of example, the dates were 1634: Jan. 9, 1635: Dec. 28, 1636: Jan. 2, 1637: Jan. 11, 1638: Dec. 31, 1639: Jan. 21 (LambCPFF 609).

There is a correlation between the mean winter temperature at Tokyo (Edo) and the Lake Suwa freezing date; the estimated temperature is 4.1oC for the 1630s and 1640s (LambCPPF 610).

—the first snow cover in Tokyo

This was January 6 (1633), December 16 (1638), February 2 (1640), November 28 (1642), and January 10 (1648). (611).

—the proportion of a cold-adapted species in pine pollen from Ozegahara, a raised bog 150 km north of Tokyo

Most of the 17th century appears to have been a bit on the warm side (Batten 18).

—tree ring data

That from Yaku Island in the south shows two sharp temperature drops in the 17th century, and from central Honshu shows slow decline in temperature during 17th century)(Batten 19, 21).

Generally speaking, the winters in central Japan were most severe in 1500-1520, 1700-10, and 1850-80, not in the period of interest to us now (LambCHMW 227).

While the Genroku (1695-6), Tenmei (1782-7) and Tempo (1833-39) famines all occurred during particularly cold periods, Japan’s population still doubled from 1600 to 1721 (Batten, 57, 59).

Pacific Ocean

The Spanish take advantage of wind (midlatitude westerlies, subtropical northeast trades) and currents (the North Pacific gyre) in the Manila-Acapulco galleon trade. Spanish archives show that the average duration of the Acapulco-Manila passage (westing made mostly around 12oN latitude) increased steadily from 80 days in 1600 to 100 in 1640 and a peak of a little over 120 days in 1655, then descended to gradually to a plateau of 90-100 days in 1690-1750. “Virtual voyage” calculations indicated that the slowing was most likely the result of a northeastward shift in the position of the “southwest monsoon trough” (the ICTZ) in June (Garcia-Herrera).


The fictional cosmic event we call the Ring of Fire replaced a six mile diameter of 1631 Thuringian air with one from 2000 West Virginia. What we will speculate about in this part is just how profound an effect this event would have had on weather (short-term) and climate (long-term), both locally and remotely.

Okay, folks, hold your hats. It’s time for our intellectual roller coaster to plunge into the abyss of chaos theory. Just be thankful that I am sparing you the mathematics that I studied, and that I am concentrating on the implications.

Let’s begin the descent gently by talking about the origin of the term “butterfly effect.” It comes from the title of a presentation by the mathematical meteorologist Edward Lorenz: “Does the flap of a butterfly’s wings in Brazil set off a tornado in Texas?”

This was a reference to his observation of chaotic behavior in his atmospheric convection model. Chaotic behavior means that a small perturbation in the initial conditions results, eventually, in a large and to some degree unpredictable divergence (“bifurcation”) in the “final” state. This chaos was not the result of randomness; Lorenz’ model was completely deterministic. However, the chaos had the appearance of randomness.

Chaotic behavior can be inherent in the actual physical system—and the existence of nonlinear relationships is necessary for chaotic behavior to emerge. It can also be an artifact of numeric evaluation of a mathematical model of a physical system. (The input data is for grid points, not spatially continuous; the evolution of the model is calculated in discrete time steps, not continuously, which means that we are working with finite differences not true derivatives; and there will be rounding errors inherent in how computers handle numbers.) Of course, it isn’t necessarily easy to separate the two!

The nonlinear dynamics of the climate system include both positive and negative feedback loops. As an example of positive feedback, increasing the surface temperature of land or water just below the freezing point results in conversion of snow and ice (high albedo) to bare earth or liquid water (low albedo), which increases absorption of solar radiation, which increases global temperature. Also, warming the air allows it to hold more moisture, and since water is a greenhouse gas that results in more absorption of solar radiation and its partial re-radiation back into space. And warming water causes it to hold less carbon dioxide, so it releases carbon dioxide (another greenhouse gas) into the atmosphere. On the other hand, an increase in the temperature will cause an increase in black body emission of infrared radiation, i.e., a loss of heat energy. An increase in atmospheric carbon dioxide will result in an increase in dissolved carbon dioxide.


The Ring of Fire is essentially what a chaos theorist would call an “initial condition perturbation.” Nonlinear dynamics has been intensely studied the last few decades, and there are several things you should know about perturbations.

First, when you perturb a nonlinear dynamic system, it initially evolves toward what the mathematicians call an “attractor.” This is the particular subset of all the possible values of all the possible variables that the system, by virtue of its dynamics, prefers to be in, i.e., gravitates toward. There are point attractors (stable states), limit cycle attractors (periodic states), and “strange” or “chaotic” attractors (which obviously are the ones that exhibit chaotic behavior). (These terms are defined by reference to something called “phase space,” but we don’t need to talk about that . . . .)

Second, depending on what part of the “attractor” the system gravitates to after the perturbation, the perturbation may grow exponentially, grow slowly, remain static or even dissipate. This can be seen with the Lorenz 1963 climate model that led Lorenz to his discovery of chaos (Kalnay).

Third, in complex nonlinear dynamic systems, it is typical for the exponential growth to reach a saturation point and level out. That doesn’t mean that the system becomes static, just that the “swings” stop getting larger and larger. One set of nonlinearities creates the exponential growth, then another set kicks in and curbs it (Toth 3298). Bear in mind, the system remains perturbed; the “weather” is still different.

Fourth, the mere size of the perturbation isn’t necessarily important. It appears that small perturbations may have the same exponential growth rate and same saturation levels as large ones; being smaller they just take a little longer to reach that level (Lopez Fig. 1). However, there is some scholarship suggesting that small amplitude perturbations are more likely to grow at a constant, not exponential, rate (Noone 8).

Fifth, it does matter whether the perturbation is a random one. A random perturbation is more likely to be inconsistent with the flow regimes established by the underlying physics, and dampen out rapidly: “purely random perturbations yield unbalanced flow structures and lead to the perturbation energy being dissipated as gravity waves during the initial time steps.” (Magnusson2002). Yes, I’ll take his word for it.

That’s relevant for the Ring of Fire, because let’s face it, the RoF rather haphazardly dumped matter and energy into one small segment of the world. While the masses may be roughly equal, they aren’t identical in toto, and certainly not in chemical composition. And the heat and pressure-based energies of the old and new hemispheres are certainly different. There is no real life physical process that could have caused the sudden change in temperature, pressure and atmospheric composition from what had been there an instant earlier.

But even “balanced” random perturbations will decay initially, until they reach the attractor (Toth 3300; Magnusson2008).


Annann and Connolley ran two runs of the 64 bit version of the HADAM3 model (this is the Hadley atmosphere model, with a horizontal resolution of 3.75×2.5o in longitude x latitude, a vertical resolution of 19 altitudes, and a standard timestep of 30 minutes) (Wikipedia/HadCM3). The two runs differed in that the pressure in a single grid point (the authors think it was somewhere in the Arctic) was changed by just one part in 1015. They then calculated the root-mean-square of the differences in pressure between the two runs across the entire grid, i.e., globally.

The difference (in pascals) was below 50 up to day 10, then started climbing rapidly, flattening in the 800-1000 range at day 25. Standard weather charts only show differences of 400 pascals, so the difference was practically insignificant, on a global scale, up to around day 15.

They also plotted the location of the differences as of days 4, 15, and 31. On day 4, the differences were mostly in the tropics. By day 15, they were mostly outside the tropics, in both hemispheres. Come day 31, and the positive differences over the USA and Europe had become negative ones.

Note that these runs demonstrate both how a small perturbation can grow rapidly and how it can then reach a saturation point. The pressure perturbation in question is much smaller than what was likely caused by the RoF, but it’s also worth noting that the spatial extent of the RoF is much smaller than than a single HADAM3 “grid box,” and that pressure is more susceptible than temperature to chaotic fluctuation.


The way that weather and climate forecasters have coped with chaos is through “ensemble” forecasting. That means that they created a set (small enough to be computationally practical) of perturbed initial conditions and ran the same weather or climate model on each of them. They then based predictions on what the entire ensemble said would happen in the future, with the reliability of the prediction being considered an inverse function of the degree of divergence of the ensemble members.

It used to be that the ensemble members were simply random perturbations of the “observed” initial conditions, with the magnitude of the perturbation related to the assumed observational error. The meteorologists ran into that damping problem I mentioned, and therefore imposed dynamical constraints on the perturbations (a fancy way of saying, selecting perturbations that were more likely to have existed in the real world).

Also, in order to maximize the bang for the computational buck, they developed tricks for selecting just the perturbations that were likely to grow the fastest, and were therefore the best test of the reliability of the prediction.

The perturbations typically used by forecasters are much larger than the RoF perturbation.

Toth (3309) talks about global-scale perturbations in the range of 10-20% of the natural climate seasonal variability (rms variance). Zorita conducted two different runs (ERIK1, ERIK2) of the ECHO-G global climate model, each starting at 1000 A.D., with ERIK2 postulating colder conditions. I have not been able to ascertain the global difference in the initial condition, but in the Baltic area, at least, ERIK2 annual mean temperature was colder by 0.5oC (Hunicke 21). If difference on the global scale is the same, that’s a huge perturbation. In fact, that difference by itself adds up to a bigger perturbation energy than what could possibly be represented by the RoF. For that reason, I can’t infer from the spread of ensemble member behavior how much the RoF will disrupt the OTL climate.


Also, as perturbations go, the RoF is small potatoes, both in size and extent. While it certainly has the potential to make the climate, not just the weather, quite different from that of the old time line, it has to compete with other external influences of much greater magnitude.

The climate evolves not just on the basis of the initial state of the atmosphere and ocean, but also as a result of external “forcings”: solar radiation, volcanic eruptions, and human activity. In a “free” (unforced) simulation for 1000 “years,” the Northern Hemisphere annual mean temperature fluctuated chaotically within a range of about one degree. This is mostly attributable to the internal variability created by the nonlinear dynamics, and is “smoke without fire.” (There was actually a “base” forcing; continuously repeated annual cycles of solar radiation.)

In contrast, in a simulation forced with assumed historical variations in “effective solar constant,” carbon dioxide and methane, the model behaved quite differently, with sharp temperature drops synchronized with downward spikes (attributable to sulfur dioxide from eruptions) in the effective solar constant, and there was an upward trend broadly mirroring the increases in greenhouse gases (Von Storch).

The level of chaos in the climate system is not sufficient to obscure the seasonal cycle of temperature, which is obviously forced by the seasonal variation in solar insolation (radiation hitting us). In almost every year of record, outside the tropics, the mean temperature for January is less than the mean temperature for July. And the occasions where the two have been close—the “years without a summer”—have been synchronized with volcanic eruptions, which are another kind of external forcing.

While it is more common for chaos to disrupt the daily cycle—while temperatures usually peak in mid-afternoon in response to daytime radiation, and reach their 24-hour nadir just before sunrise as a result of nighttime cooling, it’s easy enough for a sunset warm front to upset the pattern—I suspect that there have been few months in which the average night-time temperature has been higher than the average daytime one.

A shorter averaging period will be needed to filter out chaotic fluctuations in temperature, which is a parameter directly driven by (non-chaotic) solar radiation, than that needed to address precipitation, sea level pressure and wind speed, which are heavily dependent on “turbulent atmospheric dynamics” (Zorita).

Climate is predictable because some of the major forcings (solar radiation) and components (the ocean, soil) are mostly of a slowly varying nature, and thus impart “memory.”

“Seasonal anomalies and longer-term climate anomalies tend to be controlled by other processes whose predictability is not necessarily limited to 2 weeks.” (Buontempo 41).


Now, let’s talk about the size of the perturbation created by the RoF. The air mass may differ in temperature and pressure, and certainly in composition, from the air that it replaced.

Eric has acknowledged that the Ring of Fire took place, up-time, on April 2, 2000. We also know it occurred around noon; a few minutes after the event, Frank pointed out to Mike that the sun was in the wrong place, that it should be to the south (Flint, 1632, chapter 3).

Determining what those differences are isn’t that easy. You would think that we could at least fix the characteristics of the year 2000 hemisphere. And you would be wrong. While the Mannington 8 WNW weather station was in business in 2000, the April 2000 records are not available from NCDC. March yes, May yes, April no.

According to the Farmers’ Almanac, the closest available weather station, HARRISON MARION RGN, WV, reported on that day a high of 62.6oF, a low of 48.2, an average of 55.7, a dewpoint of 45, a wind speed of 3.9 knots, and 0.01 inches precipitation (essentially, a drizzle). The mean temperature and dewpoint correspond to a relative humidity of 67%. Another nearby station, MORGANTOWN HART FIELD, WV, reported high 61, low 48.9, average 56.9, dewpoint 39.9, wind speed 4.7 knots, and precipitation 0.09 inches.

I took a look at the 24-hour graphs for the station KWVFAIRM17 in Fairmont for April 2 in 2007 – 2010. In 2010, noon temperature was almost 5oC below the peak, reached at 6 pm. In 2009, the peak was at 3 pm and the noon temperature was only about 2.5°C less. In 2008, the noon-to-peak spread was similar but the peak was at 5 pm. And 2007 was similar.

So I am going to estimate that on the day of the RoF, the high was 62°F (16.7°C) and that at the time of the RoF, the temperature had only reached about 57.6°F (14.2°C). The saturation vapor density of water is 12.15 grams/cubic meter at 14.2°C. I am estimating a relative humidity of 60%, there’s 7.3 grams/cubic meter of water in the air.

I have no idea what the air pressure was that day. There is no reference to rain on RoF day in the novel, so the pressure was probably average (29.92 inches mercury, 1013.25 mbar) or a little above.

We know even less about the conditions on May 25, 1631 near Rudolstadt, Germany. Presently, the average max and min temperatures are 13 and 4oC for April and 18 and 8 for May (, i.e., a 9 – 10 degree spread. For 1766-1850, climate reconstructions yield an average of 12.10°C (standard deviation 1.51) for May, and 15.35oC for June (s.d. 1.24). May 25 is one-third of the way from May 15 to June 15, so the mean temperature for May 25 was probably something like 13.18oC (55.7°F).

The sun in the new timeline sky is in the east, so it’s morning. Indeed, in the mid-latitudes, temperatures tending to be coolest just before sunrise and warmest at 4-6 pm. I think it reasonable to expect the mean to be hit around 9 am.

The principal atmospheric gases are nitrogen and oxygen. The trace gases that are significant from a climatic standpoint are water vapor, carbon dioxide, methane, and the nitrogen and sulfur oxides. Besides gases, the atmosphere also contains aerosols (airborne particles), including desert dust, droplets of sulfuric acid produced by the reaction of volcanic sulfur dioxide with water, carbon from smoke, and sulfates from the combustion of fossil fuel.

My summary as to the likeliest size of the perturbation caused by the Ring of Fire, as measured by meteorological variables, is in Table 3-1.

Table 3-1: The RoF Perturbation

Grantville April 2, 2000 noon hemisphere Thuringia May 25, 1631 morning hemisphere Effect
Temperature (oC) 14.2 ~13.2 +1
Pressure ? ? ?
Water (grams/meter3) 7.2 ? ?
Carbon Dioxide (ppm)* 363.8 (1997) 370 (2000) 275.7 +94
Sulfur Dioxide (ppm) 0.015 ppm** ? [check data]
Methane (ppm)* 1750.8 (1997) 723.5 +1030?
Nitrous Oxide (ppm)* 306.4 (1997) 285.1 +25?

*1631 and 1997 values (Robertson), 2000, estimated from graph for Mauna Loa:

**. 2000 sulfur dioxide concentration for Marshall county (which is near Marion). (WVDEP).

Because the RoF occurred suddenly, it’s probably most comparable to a volcanic eruption. There is no doubt that an eruption can have a profound effect on climate, as well as weather, for up to several years, at least if it injects material into the stratosphere.

So the question is, how does the RoF rate, compared to various eruptions, as a source of heat, carbon dioxide, and aerosols?

Since the diameter of the RoF hemisphere is six miles, its maximum height above the earth’s surface is three miles. The stratosphere begins at about 6-31 miles above the ground in temperate latitudes, so the immediate meteorological effect of the RoF is limited to the troposphere.

There’s something called the Volcanic Explosivity Index. It’s based on plume height and volume. The three mile height of the RoF hemisphere corresponds to a plume height of just under five kilometers; VEI 2 (“explosive”) is 1-5 km. However, the volume (236 km3) ranks as VEI 7 (“super-colossal”), 100 – 1000 km3.

However, that’s quite misleading because the temperature difference between a volcanic plume and the “background air” is likely to be much higher than that between the Grantville and Thuringian hemispheres. And the volcanic plume is going to differ in composition from “background air” more than the two hemispheres do, too.

At the temperature of the Grantville hemisphere, the heat capacity of air is 1.005 kJ/kg-oC, and the density is about 1.225 kg/m3. Since the temperature difference between the Grantville and Thuringian hemispheres is most likely one degree Celsius, that means that the heat energy introduced by the RoF is about 2.9*1011 kJ. That’s a bit less than the amount of energy released when the water in a typical thunderstorm condenses.

For a volcanic plume measured on March 19, 2002 at Miyake Island, the temperature initially was 34oC, but the temperature dropped to 20 when the plume rose to 3 km and to 19 at 6 km. Such a plume would have been carrying at least an order of magnitude more heat energy than that attributable to the air temperature difference caused by RoF.

The RoF also resulted in an injection of greenhouse gases and aerosols, but it was small.

Even if there were no sulfate aerosol already in the Thuringian air that it replaced, the effective injection was only about ten tons sulfate, as a one-shot deal. In comparison, during the Miyake 2002 eruption, that volcano was emitting 10,000 tons SO2per day, for months. Yet it had a VEI of only 2.

Thus, it seems fair to conclude that the effect of the RoF on climate is mostly likely to be smaller than even that of an eruption of VEI 2.


I consulted with climatologist James Annan (Senior Scientist, Research Institute for Global Change, Yokohama Institute for Earth Sciences) about the possible effect of the RoF (without providing the specifics of the size of the perturbation because I hadn’t yet calculated it at the time of our email exchange). Here’s his reply:

“Generally speaking, changing the initial conditions (atmosphere) will, as you say, scramble the weather. It will not affect the response to temporally-varying forcing such as solar output, or volcanoes, which probably drive a large part of the large-scale climate changes. However, it’s not obvious to me (and perhaps anyone) to what extent the observed seasonal climate variation is simply due to internal variability, versus a forced response.

“On the assumption that the important external forcing factors are greenhouse gases and solar output you could probably consider swapping the climates around from roughly consecutive years—e.g. use the seasonal means from 1632, then 1631, then 1634 . . . and so on shuffling the years around a little. The daily weather would not match in any case. If there’s a volcano (I haven’t checked) then that would have to affect the particular year of course.”

I have since checked the volcanic activity. While there was an active eruption somewhere in the world for every year of the 1630s, none of these was likely to be larger than VEI 5, and, more importantly, ice cores from Greenland and Antarctica reveal low levels of stratospheric (volcanic) aerosols (sulfates) for the 1630s (Gao).


While we know quite a bit about the changes in temperature and rainfall in the OTL 1630s and 1640s, especially in Europe, there are uncertainties as to what will happen in the NTL. Some of those uncertainties are attributable to the limitations of reconstructing climate data from historical or proxy records, and others to the effects of the RoF. Consequently, our authors do have some flexibility as to how they portray post-RoF climate, and of course, climate isn’t weather; on any given day, there can be significant variation from the climatic norm.

In the course of researching this article, I came across the interesting tidbit that “in 1371/2 there were processions for rain at Florence in December, followed by processions praying for the rain to stop in May.” (Grove 328). Hence, it is fair to say that what is certain about both climate and weather after the RoF is that people will complain about it.


Note: The bibliography for this article will be published at in the Little Ice Age Addendum.