Jean-François Gaultier was a physician in the French colony of Québec in New France from 1742 to 1756. During that period, he recorded daily readings of temperature and observations of the weather, although only the observations for 1742–46, 1747–48, and 1754 have been located to date. Daily instrumental temperature data from Québec for the 1740s provide a glimpse of climate variability in eastern North America, upstream of the North Atlantic. During the 1740s, winters appear to have been milder than during most of the twentieth century with the exception of the 1950s and early 1980s, and summers warmer than those of the twentieth century, with the exception of the 1970s and 1990s. Autumns and springs appear to have been cool relative to the twentieth century, suggesting that, while winters may have been milder, the winter season lasted longer, with a consequently shorter growing season. The cool springs and autumns, combined with warm winters and summers, give these few years in the 1740s annual average temperatures comparable to those averaged over the twentieth century; the annual average temperatures mask marked seasonal differences. There is also some evidence that the climate was drier than in recent times, with fewer precipitation days than during the 1970–2000 period.
The study of recent climate variability, on the decadal to century timescale, is of critical importance to understanding present and future climate change. Anthropogenic climate change is superimposed on a natural climate variability, which is still poorly understood on a regional scale (Mann et al. 1998). While relatively complete climate datasets exist for the last 50 years, and some even for the past 100 years (e.g., Jones et al. 1999; New et al. 2000), it is necessary to better understand the variations and trends of climate over the past 200 to 500 years in order to place in context those of the past 50 years. This is also necessary to increase our confidence in both our understanding of climate processes and in future climate predictions on the interannual to decadal scale. Historical data, particularly instrumental observations, are essential to our understanding of past climate changes. Such data can be found in the records kept by amateur meteorologists in the periods before the establishment of national meteorological agencies in the mid to late nineteenth century (e.g., Manley 1953, 1962, 1974; Legrand and LeGoff 1992; Slonosky et al. 2001b). These data can give valuable insights into the variability and behavior of climate during the period before the impact of industrialization, help evaluate the natural variability of climate, and place recent climate change into a longer-term perspective (Folland et al. 2001).
Current ideas of climatic change of the past 500 years are dominated by the concept of the “Little Ice Age,” a period of worldwide glacier advances lasting from the thirteenth to the nineteenth centuries (Grove 2001). The evolution of climate during this period is somewhat more complex, with cooler temperatures particularly in the seventeenth and nineteenth centuries, and subsequent recovery to warmer conditions during the twentieth century, along with anthropogenic climatic change during the late twentieth century (Jones 1998; Mann et al. 1998; Crowley 2000). Recent advances in the field of historical climatology have added considerably to this picture, building a more complex and detailed picture of the climatic changes that have occurred over the past few centuries (e.g., Wilson 1983; Harrington 1992; Legrand and LeGoff 1992; Pfister et al. 1994; Mann et al. 1995; Luterbacher et al. 1999, 2002; Moberg et al. 2000; Auer et al. 2001; Demarée and Ogilvie 2001; Jónsson and GarUarsson 2001; van Engelen et al. 2001). Considerable interannual climatic variability existed during all these periods, with alternating years and even alternating seasons of extreme wet or dry conditions (Slonosky 2002), and months or seasons of extreme warmth and cold linked to variability of the atmospheric circulation (Luterbacher et al. 2000; Jacobeit et al. 2001; Shindell et al. 2001; Slonosky et al. 2001a,b) in the seventeenth and eighteenth centuries. The use of early instrumental meteorological records has also led to several extremely important developments in the past few years, including the extension of an index of the North Atlantic Oscillation (NAO)—one of the most important modes of atmospheric circulation in the Northern Hemisphere—on a monthly scale back to 1822 using instrumental data (Jones et al. 1997), and back to 1675 on a monthly scale and 1500 on a seasonal scale using historical documentary data (Luterbacher et al. 1999, 2002). These reconstructions enable the study of climatic change and climate variability to be extended over several hundreds of years, and the dynamics of the atmospheric circulation over these periods to be examined in detail (Jacobeit et al. 2001; Slonosky et al. 2001a).
The years of interest to this study, the 1740s and 1750s, were during the cool centuries of the Little Ice Age, but after the period of very low solar activity known as the Late Maunder Minimum (LMM) from 1675 to 1715, during which the climate was particularly severe in Europe (e.g., Pfister et al. 1994; Legrand and LeGoff 1992; Slonosky et al. 2001b; Luterbacher et al. 2002). The seventeenth century was the coldest of the past millenium, followed by the nineteenth (Jones et al. 2001). During the eighteenth century, Northern Hemisphere temperatures were on the order of 0.3°C lower than those of the late twentieth century, the 1750s showing a relatively warm peak in the Northern Hemisphere temperature records, compared to lower temperatures in the 1680s and 1690s and 1800–50 (Jones 1998; Mann et al. 1998; Jones et al. 2001).
Although it is not easy, and sometimes not possible, to adjust historical data to conform to modern standards, some adjustments can be made to account for the likely impacts of more primitive instruments and different observing practices than those in use today. The keen interest of the early amateur observers, their intelligent use of their instruments and interpretation of their observations can sometimes compensate a great deal for the deficiencies in the instruments (Manley 1953, 1962, 1974; Legrand and LeGoff 1992; Slonosky et al. 2001b; Slonosky 2002). Despite the limitations of historical instrumental records, even after calibration to modern standards, some conclusions may be drawn as to the general character of the seasons and the interannual variability of the period for which the data exist.
This paper examines historical climate data that has been largely forgotten by the climatological community, recently found in the Paris Observatory in France. Records of daily temperature readings, weather observations, and observations of the state of agriculture and vegetation were kept by Dr. Jean-François Gaultier, who was sent as the king's physician to the French colony at Québec for the period 1742–56. These are the earliest known instrumental meteorological observations for Canada, and among the earliest of the North American continent (Baron 1992). The instrumental observations and adjustments to modern observing standards are discussed in section 2. In section 3, a description of the climate during the 1740s is given, based on both the instrumental observations and Gaultier's descriptive and phenological observations. A comparison between the climate of the 1740s and that of Québec City today is given in section 4, followed by a discussion of Gaultier's data in the context of other historical and paleoclimatic observations, as well as reconstructions of atmospheric dynamics, in the final section.
2. The meteorological observations of Jean-François Gaultier
Jean-François Gaultier (1708–56) was a physician educated in Paris, and sent to the city of Québec in 1742 as the king's physician to the colony in New France. He was a corresponding member of the Académie Royale des Sciences, corresponding with H. Duhamel du Monceau and later with R.A.F. Réaumur on the botany of the New World, its climate and on his thermometric experiments. His daily thermometric measurements, precipitation-type observations, wind direction, and weather descriptions were printed in the Mémoires de l'Académie Royale des Sciences (Fig. 1; Duhamel du Monceau 1744, 1745, 1746, 1747). Manuscript observation diaries also exist at the Paris Observatory for 1744/45, 1747/48, and at the Houghton Library of Harvard University for 1754 (Gaultier 1745, 1748, 1755). Gaultier recorded monthly summaries of the weather, the state of the colony's agriculture, which was highly dependent on the weather for crop growth and favorable planting and harvesting conditions, and the public health of the colony. He remained in Québec City until his death in 1756. His scientific contributions, particularly his accomplishments in the fields of botany and meteorology, and the quality of his phenological observations have been recognized by historians (Vallée 1930; Wien 1990). Evidence suggests that he kept meteorological records during this entire period, although several years of observations appear to have been unfortunately lost.
The thermometer readings were almost always taken in morning, between 0700 and 0800 local time (LT) and occasionally also in the afternoon between 1400 and 1500 LT (Fig. 1). The morning readings are quite consistent, with only 1.6% of observations missing. However, the afternoon readings were only taken occasionally, and seem to have been recorded particularly when the afternoon temperature was warm for the time of year. Most of the afternoon temperatures are recorded on mild winter days, in spring, and in summer.
The thermometer readings have been converted from Gaultier's scale, which was based upon that of Delisle for 1742–48 (Duhamel du Monceau 1744). Gaultier's early thermometer was made of mercury and his scale designated 0 as the freezing point of water, but put 100 at the boiling point of alcohol, or about 80°C (Middleton 1966); these values were converted into a centigrade scale. By 1754 however, Gaultier seems to have acquired a centigrade thermometer. There is a bias against extremely cold days, as the mercury in Gaultier's early thermometer contracted into the bulb at about −23°C, although Gaultier still attempted to quantify the degree of contraction. For the purposes of this study, an approximation has been made by replacing days on which Gaultier's thermometer fell to below −23°C with an extreme value calculated as the modern daily mean minimum value minus two standard deviations. The daily mean minus two standard deviations was found to provide the best estimated value consistent with the evolution of the daily temperature during these cold spells, producing no sudden jumps in the record. Eighteen days were replaced in this manner during 1742/43, four during 1743/44, two during 1744/45 and 1745/46, and six during 1774/48; none were replaced in 1754. It was decided to use these estimated values to replace the temperature values for these days of extreme cold in preference to coding the values as missing, in order to bias the data as little as possible; coding these days as missing would have introduced a warm bias in the data. (This problem faced by Gaultier in Canada as well as by observers in parts of Russia led to the development and calibration of new thermometers with an extended scale.)
Figures 2a and 2b show the daily morning and afternoon readings, together with the mean plus and minus one and two standard deviations from the modern Québec minimum (Fig. 2a) and maximum (Fig. 2b) temperatures, adjusted for Québec City. The current observation location is at Québec City airport, a small regional airport. See Vincent et al. (2002) for details of the homogenization of daily temperatures. The adjustment factor determined by Vincent to render the Québec City series homogenous, given several site relocations (including one from the city center to the airport), ranges from −2.7°C in February to −1.3°C in May for maximum temperature, no adjustments were needed for minimum temperature (L. Vincent 2001, personal communication). Gaultier's values generally fall within the expected range of temperature values. As the minimum temperature usually occurs near sunrise under normal clear sky radiative conditions, Gaultier's fixed-hour observations are taken close to the time the minimum temperature occurs in winter, but are a few hours later in summer. His summer morning values are thus higher than the minimum Québec temperatures in summer.
Figure 3 shows the estimated maximum and minimum temperatures estimated from Gaultier's readings. These estimates were made by calculating the difference between the 0800 LT and minimum temperature for each day of the year, and the 1500 LT and maximum temperature for each day of the year, using hourly data from the Québec City airport from 1970–2000. The 0800 and 1500 LT temperatures were chosen to allow a maximum lag between the minimum/maximum and the hour of observation, and to take into account the difference between solar time (by which Gaultier would have reckoned) and standard time, upon which the modern observations are based. Even with this adjustment, the minimum temperatures were generally high, being close to one standard deviation above the modern mean minimum temperature, especially in winter.
Almost no information is given as to the placement of Gaultier's thermometer, except that it was placed with a north and northwest exposure, these being the directions of the coldest winds (Gaultier 1748). Given the practices of the time (e.g., Manley 1953, 1962; Wilson 1983; Legrand and LeGoff 1992; Chenoweth 1993; Slonosky et al. 2001b), it seems likely that the thermometer was either hung on an outside wall of Gaultier's house, or placed near a window of an unheated room. As the thermometer regularly gave readings below −20°C, and as the socioeconomic conditions of life in New France made the luxury of an unused room improbable (with such cold temperatures, it would be impossible to carry out any daily tasks in an unheated room in winter, or even to record the temperature readings, as water and ink would have frozen solid), exposure on an outside north-facing wall, probably outside a window where the thermometer readings would be visible from the inside, seems to be the most likely exposure to the author. According to Wilson (1983), Chenoweth (1993), and Parker (1994), the unscreened north-wall exposure may lead to biases due to exposure, ventilation, and heat retention by the building. Parker (1994) concluded that north-wall exposures led an enhanced diurnal cycle compared to the observations taken within a Stevenson screen, but the differences between a standard Stevenson screen and a north-wall exposure are very site-dependent. Biases for a north-wall exposure estimated by the detailed studies of Wilson (1983) and Chenoweth (1993) due to ventilation and exposure to shortwave radiation are based on relatively short studies, some only a few months long, and range between 0.5° and 2.6°C for the minimum temperature and from 0° to −1.9°C for the maximum temperature (Chenoweth 1993), the overall bias for monthly mean temperature ranging from 0.2° to 0.6°C (Wilson 1983).
For Gaultier's readings, given the lack of information concerning the building, the degree of urbanization of the site and the probable exposure to shortwave radiation, it is the author's considered opinion that the most serious problem is likely to have been thermal lag in the transfer of heat from the house (when heated) to the thermometer, as the difference between the indoor and outdoor temperature is likely to have been at least 25°C on the coldest winter days, for the socioeconomic reasons outlined earlier. It is assumed the indoor temperature would have to be, at the minimum, near zero, and as we have seen, Gaultier's thermometer regularly went down to at least −23°C. This effect would be most important for the coldest observations, those of winter mornings. Accordingly, additional adjustment factors were determined by adjusting the minimum values to fit with the weather descriptions given by Gaultier. This was done by ensuring that when Gaultier described frost or snow, the minimum temperature was at least 0°C, and conversely, if the day was described as frost-free, ensuring the minimum value was above 0°C. The adjustments determined for the minimum temperatures were 0.5°C for 1–15 April and 15 September–1 October, 1°C for 15–31 March and 1–15 October, 1.5°C from 1–15 March and 15–31 October, and 2°C from 1 November–15 March. No additional adjustments were applied during the summer months as it is assumed the house would not have been heated, and the maximum temperatures are unaffected. All these adjustment factors were filtered using a 20-point Gaussian filter to provide a smoothed annual cycle in the adjustment factors.
A check on these estimated values was made by comparing the length of the frost-free season as determined by Gaultier's weather descriptions to the length of the frost-free season defined by the number of days between the last frost (last day with a minimum temperature at or below 0°C) in spring and the first frost in autumn; these are shown in Table 1.
The only major discrepancy is in 1745, when Gaultier describes an unusual frost having occurred during the night in late August but with a minimum temperature still above 0°C (Gaultier 1745); this is possible if strong radiative cooling occurred near the ground, but the air temperature a meter or so above the ground was still above freezing. It is also possible that Gaultier is reporting frostlike damage, rather than a “killing frost” associated with below-zero temperatures (C. Mock 2002, personal communication), and so this early frost date was left as anomalous. The next frost day described by Gaultier in 1745 coincides well with the date of the first below-zero minimum temperature.
3. Warm summers, variable winters: Descriptions of the climate of the 1740s
It is always difficult to judge weather descriptions, as they are subjective measures and depend on the observer's personal prior experience of weather and climate. As a newcomer to the colony from France, Gaultier's perceptions of the weather extremes may be exaggerated, although he himself was aware of this possibility and took care to also record the remarks of inhabitants who were born in the colony. Even these may be flawed, however, and so more objective descriptions, such as the state of the snow cover, the arrival of migratory birds or the appearance of rare species and the dates of harvest of various fruits and grains are important “independent” descriptors. Gaultier's detailed descriptions are given in the following paragraphs, and a summary of the seasons for 1742–47 is given in Table 2.
Although there are no instrumental observations before November 1742, the summer of 1742 was described as very hot and dry. In January 1743, there were sufficient caribou near Québec City to form a caribou hunt; this was a rare occurrence according to Gaultier, as only in exceptionally cold winters did caribou migrate as far south as Québec City. By February 1743, there were 8–10 m (25–30 French ft) of snow and snow drifts in some places. The summer of 1743 was hot during June and July.
An ice storm on 22 and 23 December 1743 left 8 cm (3 French in.) of ice on the roofs and walls of houses. A warm spell in the middle of January 1744 prompted buds to appear on some trees, but this was followed by a very cold spell “the worst in living memory; in less than 10 minutes, nose, ears and lips froze, eyelids froze shut and it was very painful to be outside” (Duhamel du Monceau 1745). These alternating warm and cold spells, with associated freeze–thaw cycles, continued through February. The rest of the spring continued mild, and the sowing was finished by May 11, one month earlier than in 1743. The summer of 1744 was favorable for agriculture, with fair weather throughout the summer.
A warm spell in November provided a brief return to summerlike conditions, with the fields still green, and the weather “infinitely nicer than usual for the season” (Gaultier 1745). By 17 December, however, it had become sufficiently cold that the nearby St. Charles River had an ice cover thick enough to sustain the weight of all kinds of vehicles. By March, the winter was considered to be the “mildest in living memory” (Gaultier 1745). June 1745 was described as an “extremely hot month,” although the strawberry harvest was not ripe until 22 June, 10 days later than in 1744. Gaultier describes June and July of 1745 as “excessively hot, but with cool nights … and the countryside beautiful and merry” (Gaultier 1745). By the end of August, however, the temperature had cooled down to the point of hoar frost on 30 and 31 August (although this is not reflected in the temperature measurements).
November and December 1745 were mild. It froze hard during the middle of December, but another warm spell at the end of December melted most of the river ice near the edge of the St. Lawrence River. The weather remained very variable, with alternating cold and warm spells; some days the temperature was above freezing, others it was so cold the mercury condensed into the bulb of the thermometer (i.e., colder than −23°C). These occasional hard frosts but frequent thaws led Gaultier to remark that “for Canada, it is very mild” (Duhamel du Monceau 1747). By March, it was stated once again, for the second year running, that “the oldest inhabitants of the colony don't remember such a mild winter, so mild that the St. Lawrence didn't even freeze over at Québec.” July and August 1746 are described as very hot, with so little rain that the rivers, wells, and other water sources dried up, and the leaves began to wither on the trees.
e. Winter 1746/47
A brief description of 1746/47 is found at the beginning of the notes of 1747, and reads as follows:
“The winter of 1747 was terrible, long and bitterly cold. All the lakes, rivers and even the St. Lawrence froze thick, so that carts and sleds could pass over. There was a bridge of ice on the St. Lawrence all the way to Montreal. The spring and summer were favourable for vegetation, the snowmelt started early. The summer was hot; this excessive heat made the crops grow quickly, but the heat was tempered by rain storms. The heat scorched the wheat and destroyed the harvest in a few areas. In general, however, 1747 was very abundant for all kinds of production.” (Gaultier 1748).
December of 1747 was very cold, and by the end of December the river ice was thick enough to bear loaded carts everywhere but on the southern edge of the St. Lawrence River. January brought severe cold weather and harsh blowing snow, which made it impossible to go out even on sunny days. Gaultier estimated that 130 cm (4 French ft) of snow fell in January, and the rough weather continued in February and March. June 1748 was hot, with such great heat that many plants and leaves dried out in the heat. The first half of July was excessively hot; “perhaps such great heat has never been felt before in Canada” (Gaultier 1748). The weather cooled off to such an extent that ice patches were seen on 20 September.
4. Climate change? A comparison with modern data
a. Temperature and frost days
From Figs. 2 and 3 the winter of 1742/43 is seen to have been cold. The winter of 1747/48 was also relatively cold, and 1746/47, although no thermometer readings have been located, was described as a “terrible winter, both for the degree of cold and its length” (Gaultier 1748). Even with the adjustment factors described in section 2, the summers appear to have been somewhat warmer, on the whole, than the average over the past century. Occasional comments as to the state of the harvest or other agricultural phenomena tend to confirm the warmer, or possibly drier, summers: in June 1744, the strawberry crop was ripe by 12 June (in modern times, the strawberry crop of Québec City is often not ripe until the end of June or beginning of July at the earliest; mid to late July is more common), and in 1743 the heat caused the apples to fall off the trees prematurely in July. With such a small sample, it is impossible to tell if this is evidence of a climatic change or normal interannual variability. However, both the written descriptions, the instrumental data and the phenomenological descriptions point to relatively warm summers and variable winters; the winters of 1742/43 and 1746/47 were both cold, but the winters of 1744/45 and 1745/46 were mild, with frequent thaws and warm spells throughout the winter. Winter seems to have started later, with mild autumns, but to have continued longer into the spring, with snow a common occurrence as late as May. The t tests show significantly warmer Januarys and cooler Mays in Gaultier's records compared to the 1970–2000 period (see appendix Table A1).
The morning temperatures for 1742–54 cluster near the same values as for the modern period, although in winter and autumn they tend to be nearer the upper values of the modern period; January of 1754 is warmer than any during the 1970–2000 period. The springs of 1742–54 were cooler than during the modern period, especially March 1748 and April of 1742 and 1745, suggesting that the winters lasted longer, even if they were not as cold. The summers were quite variable, with June of 1748 being colder than any of the past 30 yr, but August of 1744 warmer than any of modern times.
The 1500 LT temperatures are somewhat higher than during modern times, especially in autumn and winter but as has been noted in section 2, there is a bias against colder days, as the afternoon temperatures seem to have been recorded when the weather was unusually mild for the time of year, especially in winter. The summer monthly afternoon temperatures of 1743–54 fall within the general range of those for 1970–2000; although July of 1748 stands out as being exceptionally hot.
The number of frost days, or days when the temperature went below zero, was determined for each month from Gaultier's weather descriptions, and are shown in Fig. 6. These agree well with the tendencies seen in the minimum temperatures; the winters of the 1740s had fewer frost days than in modern times, and Gaultier describes several of the winters having pronounced freeze–thaw cycles, a characteristic of modern winters in Montreal, some 250 km to the southwest of Québec City. The months from October to May have fewer frost days during Gaultier's period than during 1960–95, those of November, April, and May having significantly fewer, as determined using t tests (see appendix Table A2).
Although the only measurements taken by Gaultier were the readings of his thermometer, he kept detailed descriptions of each day's weather, and from these it is possible to obtain a number of counts: the number of days each month with rain, snow, fog, hail, thunder, freezing rain, blowing snow, frost, and snow on ground. Many of these are also available for Québec City for the 1960–95 period (Mekis and Hogg 1999).
A comparison of these climate indicators shows that for almost all counts relating to precipitation, the number of days with precipitation-related events, including rain, snow, and fog, was significantly lower for all months during the 1740s than during 1960–95 (Fig. 7). Some of this deficiency may be due to underreporting by Gaultier. Gaultier is unlikely to have noticed trace precipitation events during the night, and may not have recorded all trace or light-precipitation events during the day, despite his frequent recordings of “petite pluie” (small rain). As trace events were not counted as precipitation days in the modern data (E. Mekis 2001, personal communication), this possible deficiency is not likely to be larger than the difference observed between the two periods. The possible differences in observing measures and methods of counting precipitation days make any conclusions about changes in precipitation-day frequencies less robust.
With these limitations in mind, the precipitation-day counts do indicate less precipitation, both rain and snow, during the 1740s and 1754. The t tests (see appendix Table A2) show these differences to be significant for all months. There are also statistically significant fewer fog days for all months in the 1740s and 1754, fewer days with snow on ground for all months from December to April, fewer days with freezing rain for all months from November to April, and fewer days on which thunder was heard for all months from April to September (see appendix Table A2). The ratio of rain to snow was slightly higher in the 1740s and 1754 for January and February, but lower in the other months. As more precipitation fell as rain in January and February, this would indicate warmer winters, but as the reverse is true for the other months, notably spring and autumn, this is a further indication that the other seasons were slightly cooler, with more snow than in modern times.
By far the most prominent wind directions recorded were southwest (48.5%) and northeast (34.4%), reflecting the orientation of the St. Lawrence valley (Table 3). The St. Lawrence valley lies between the Appalachian highlands to the south and the Laurentian highlands to the north. The orography of these higher-elevation areas and the St. Lawrence lowlands is sufficient to channel the wind flow of southern Québec in a predominantly southwest to northeast direction. The weather systems and storms tracks of this region tend to track up the St. Lawrence River towards the Gulf of St. Lawrence and Newfoundland to the northeast, before heading across the Atlantic toward the Icelandic low. The dominance of winds from the southwest and northeast is partly an effect of this funnelling of the flow through the St. Lawrence valley (Powe 1968). The next most common wind direction is northwest (11.2%), and is the one Gaultier associates with the coldest weather, attributing this to “the winds [from the north and northwest] which pass over Hudson's Bay and over some mountains [the Laurentians] which are always covered in snow and ice, and always have been.” (Gaultier 1748). This was a misperception on the part of Gaultier; evidence from the Hudson's Bay Company archives (Catchpole et al. 1976; Ball 1992) indicates that Hudson's Bay and the Laurentians were not perpetually covered in snow and ice in the eighteenth century, and they are certainly snow- and ice-free during the summer months in modern times. There are very few winds recorded from other directions.
d. Seasonal means and interannual variability
Figure 8 shows the seasonal means and standard deviations of Gaultier's observations, compared to 6-yr running means and standard deviations for the period 1895–1995 at Québec. (All series are shown for the minimum temperature.) From Fig. 8a, it can be seen that the winter average temperature during this brief period was indeed warm relative to the twentieth century; only the winters of the 1950s and the early 1980s were as warm or warmer than the 1740s. The winters of the late 1980s and 1990s were cooler. The summers were also warm relative to the twentieth century, with only the 1970s and the 1990s showing summer conditions as warm as or warmer than the 1740s. The springs and autumns, on the other hand, are cool relative to the twentieth century, and were colder than springs and autumns have been since the 1950s. As has been discussed earlier, the winter season, while milder, may have been prolonged through cooler autumns and springs. The interannual variability, as seen from the few years of Gaultier's observations and the 6-yr running means during the twentieth century, has remained largely the same, except for a slightly lower interannual variability during the autumns of the 1740s.
5. Discussion and conclusions
Assessing the quality of the data is always a problem when using historical data, especially if little is known about the instruments and their exposure. While there are few, if any, published paleoclimatic series of sufficient temporal resolution in the St. Lawrence valley region with which it is possible to directly compare Gaultier's data, there exist several studies of historical documentary data and paleoclimatic data from eastern and northern North America. Evidence from documentary sources from the Hudson's Bay Company archives indicate that the autumns of the 1740s near southern James Bay were warmer, with a later date of river freeze-up than during the nineteenth century, and that the springs were cool, with a later date of river ice breakup in the spring (Catchpole et al. 1992). From the Hudson's Bay Company archive data, 1743/44 is classified as a very mild winter at both Churchill (on the western shore of Hudson's Bay) and at York Factory (on the southwestern shore of James Bay); 1740/41, 1747/48, and 1748/49 are classified as severe in both places (Ball 1992); which agrees with Gauthier's descriptive observations at Québec City as far as 1747/48 is concerned (Gaultier 1748). Gaultier's very mild winter of 1745/46 and very cold winter of 1746/47 do not seem to have been remarkable further north and west. There are no meteorological observations available at Québec City for the winter of 1746/47, and the perception of cold may have been slightly exaggerated in comparison with the mild winters of the previous few years. These documentary data from the Hudson's Bay Company also indicate that the 1740s, 1750s, and 1760s were dry at York Factory (Ball 1992). Documentary data from New England, in the northeastern United States, also indicate the 1740s were dry, particularly during the growing season (Baron 1992). Analysis of the frost-free season for New England shows that the 1740s were the most variable decade of the last 250 years in terms of the number of frost-free days (Baron et al. 1984).
Dendroclimatic observations also show that the summers of the 1740s were relatively warm and dry; tree-ring composites over the entire northern North America indicate that the 1740s were as warm as the 1950s (d'Arrigo and Jacoby 1992). D'Arrigo et al. (1999) describe similar results for annual temperature over the Northern Hemisphere; the 1730s and 1750s were warm, as warm as the 1951–80 period, while the 1740s were slightly cooler. The year 1740 is recognized as being one of the coldest in Europe in over 300 years (Luterbacher et al. 2002), and both 1740 and 1742 rank among the 30 coldest Northern Hemisphere summers of the past 600 years, based on tree-ring data (Briffa et al. 1998), in contrast to Gaultier's description of 1742 as a warm summer. Tree-ring evidence from northern Québec and the Hudson's Bay region shows that this was a period of general warmth between two severe phases of the Little Ice Age in eastern North America, the first during the late seventeenth century and the second after 1815 (Guiot 1985; Scott et al. 1988; S. Payette 2001, personal communication). Borehole evidence of annual average temperature also suggests that the Little Ice Age cold signal near the Québec City region and Atlantic Canada is less pronounced than elsewhere in eastern and central Canada (Beltrami 1992). Warm sea surface temperatures in the Labrador Sea during the Little Ice Age period (Keigwin and Pickart 1999) may also have led to warmer conditions in Atlantic Canada and possibly also Québec. Tree-ring reconstructions of precipitation in the eastern United States also indicate the 1740s were dry in the regions bordering on the province of Québec (Cook et al. 1992). Overpeck et al. (1997) found that many Arctic sites were warmer during the 1700–1820 period than during the nineteenth century, while further west, glacial evidence from the Rockies show that the most extensive glacial advances of the Holocene were during the eighteenth and nineteenth centuries, and tree-ring densities suggest that the early eighteenth century was particularly cold (Luckman 1996).
Gaultier's instrumental and descriptive observations show that the climate of these few years of 1742–48 and 1754 appear to fall within the generally expected range of climate variability for the twentieth century. There are some indications that the winters in particular were milder, from the temperature observations, the number of frost days, and the ratio of rain to snow. However, it is difficult to assess how well these few years of data reflect the climate of the decade. One winter for which the instrumental observations are lost (1746/47) was described as extremely harsh, and Gaultier describes the mild winters as unusual, and the hot summer of 1748 as “possibly the warmest ever in Canada” (Gaultier 1748). These comments suggest that the warmth of these years may have been unusual.
The counts of rain and snow days are also much below the modern period. While some portion of these low values may be due to underreporting and other uncertainties in Gaultier's observations, it seems likely, based on the supplementary phenological observations summarized in section 3 and by the historical and paleoclimatic studies cited in the previous paragraph, that some part of these low counts are due to an actual reduction in precipitation. Studies of historical data in France point to the 1740s as being a dry period, and show a consistent upward trend in precipitation from the seventeenth century to the present (Slonosky 2002). Northern Hemisphere precipitation data of the twentieth century indicate an increase over the mid and high latitudes of the Northern Hemisphere of a rate on the order of 7%–12% century−1 (Folland et al. 2001). Several authors have commented on the more continental climate of the seventeenth and eighteenth centuries in Europe, compared to a milder, more maritime climate in the twentieth century, based on instrumental (Jacobeit et al. 2001) and documentary (Ogilvie and Jónsson 2001) historical data.
From the evidence of milder, or at least more variable, winters and reduced precipitation, it may be possible to infer some changes in the atmospheric circulation. It seems likely that there was increased meridional flow in winter, as the descriptions of intense cold alternating with thaws are characteristic of winters with increased meridional flow in this area, bringing alternating subtropical and polar air masses into the region. Reconstructions of the NAO by Luterbacher et al. (1999, 2002) show the NAO to have been low or negative (indicating meridional flow) from 1740 to 1742, mostly positive (indicating westerly flow) during 1743 and 1744 and then mostly low or negative again from 1745 until the summer of 1748. Jönsson and Fortuniak (1995) reported higher frequencies of easterly winds for northwestern Europe during the 1740s, and the proxy reconstruction of the NAO by Cook et al. (1998) also indicate generally low values of the NAO during the 1740s. The model study of Shindell et al. (2001) shows an increase of 0.34°C in global average temperatures between 1680 and 1780 based upon model results simulating the expected climate change due to changes in insolation and the modeled changes in atmospheric circulation, particularly the Arctic Oscillation (AO, similar in concept to the NAO in that it represents an important mode of atmospheric variability over the Northern Hemisphere and is positively linked to the strength of the zonal flow) during the 1680–1760 period; the 1740s fall in the middle of this modeled transition to more zonal circulation and warming over parts of eastern North America. However, neither the NAO nor the AO have any correlation with temperature over southern Québec, at least in modern times (Hurrell 1996; Thompson and Wallace 1998; Slonosky and Yiou 2002), although the NAO reconstructions do provide some indications as to the strength of the westerlies, albeit farther east, over the North Atlantic/European sector. The temperature observations are also consistent with the possibility that the winter polar jet stream was positioned farther north, allowing Québec City to come under the influence of warm subtropical air masses more frequently; this would also account for the more pronounced winter freeze–thaw cycles, although not for the reduced precipitation. Hot and dry summers are also characteristic of increased meridional and blocking flow, with fewer large-scale cyclones passing regularly through the region from the Great Lakes to the Atlantic Ocean.
These early instrumental data from Québec City add to the growing picture of complex spatial and temporal climatic variability on regional scales during the Little Ice Age. The seventeenth century is classified as the coldest century of the past millennium over the Northern Hemisphere, followed by the nineteenth century (Jones 1998). From Northern Hemisphere temperature reconstructions, the eighteenth century appears to have been milder, although still cooler than the twentieth (Jones 1998; Jones et al. 2001). In this region, parts of the eighteenth century were as warm as some of the warmest years of the twentieth, at least for the winter and summer seasons. The spring and autumn seasons were cold relative to twentieth century averages, bringing the annual average temperatures of these years of the 1740s down to near the annual average of the twentieth century. All seasons were significantly drier than the twentieth century. It is clear that this period of the Little Ice Age in eastern North America was not one of uniform coldness, and that considerable interannual variability existed during this period.
These early observations are of great importance, as they are among the first instrumental meteorological records from North America, and so can provide a window on the climate and climatic variability at a daily resolution for this site in eastern North America. Other archival instrumental records are also available for the St. Lawrence valley, covering most of the nineteenth century, which may be able to provide longer-term records, also at a monthly or daily resolution, of climate change and variability in this region. These records, like all reliable historical instrumental records, are of vital importance to the understanding of climate variability and change on decadal to century timescales, as well as to placing in context the climatic changes of the past few decades. Every effort should be made to ensure that these data are not lost, and that their potential be realized.
Many thanks are due to the librarians and archivists of l'Observatoire de Paris and to Thomas Wien for their help and advice on the manuscripts of Jean-François Gaultier. The meteorological data for 1754 were obtained from the Houghton Library, Harvard University. The author is grateful to Lucie Vincent, Éva Mekis, Xuebin Zhang, Francis Zwiers, and Jacques Derome for many interesting discussions. Lucie Vincent and Éva Mekis provided the twentieth century temperature and precipitation data for Québec City. The apt and insightful comments of Mike Mann, Cary Mock, and Daniel Druckenbrod have considerably improved this paper.
Changing Climate: Differences between Eighteenth- and Twentieth-Century Québec City Climate
The following tables compare climate data for 1742–54 and the late twentieth century.
Current affiliation: Ouranos Consortium for Regional Climate Change and Adaptation, Department of Atmospheric and Oceanic Sciences, McGill University, Montreal, Quebec, Canada
Corresponding author address: Dr. Victoria Slonosky, Ouranos Consortium for Regional Climate Change and Adaptation, Department of Atmospheric and Oceanic Sciences, McGill University, 550 Sherbrooke St. W, 19th floor, West Tower, Montreal QC H3A 1B9, Canada. Email: firstname.lastname@example.org