SEARCH

SEARCH BY CITATION

Keywords:

  • historical climatology;
  • daily temperature;
  • early meteorological instruments;
  • 18th and 19th century climate;
  • Quebec;
  • Canada

Abstract

  1. Top of page
  2. Abstract
  3. Dataset
  4. Introduction
  5. 1 Thermometers and exposures
  6. 2 Description of historical observations
  7. 3 Comparisons between historical and modern observations
  8. 4 Data set quality and use
  9. Acknowledgements
  10. References
  11. Supporting Information

Daily observations of weather and climate for the province of Québec, Canada, start in the 18th century and continue to the present day. Daily temperature observations from 12 observers ranging from 1742 to 1873 are described here. The frequency distributions of the temperature observations from each of the historical weather journals are examined for data quality and consistency. Adjustments for differing types of exposures, particularly north wall exposures, are developed. It is shown that examination of the daily data distribution can be used to infer information concerning the instruments used and likely exposure in the absence of metadata. Comparisons of the relative frequency distributions of historical and modern hourly observations are used to assess the reliability of the daily historical temperature data, and are able to detect problems with instrument exposure or sampling biases. Historical observations of temperature from the 18th and 19th centuries are shown to be comparable to modern temperature data. These daily observations will be used in further studies to analyse changes in climate and extreme conditions on a decadal to centennial time frame, and will form part of international data sets for the reconstruction and analysis of past climate events.


Introduction

  1. Top of page
  2. Abstract
  3. Dataset
  4. Introduction
  5. 1 Thermometers and exposures
  6. 2 Description of historical observations
  7. 3 Comparisons between historical and modern observations
  8. 4 Data set quality and use
  9. Acknowledgements
  10. References
  11. Supporting Information

This article presents climatic data for the St. Lawrence Valley region of southern Quebec transcribed from manuscript and early printed copies of weather registers located in several archives in Canada and France. The transcription from paper and image format to numerical digital formats was largely undertaken by a web-based volunteer project.

The earliest continuous daily instrumental observations for Canada discovered to date are those of Jean-Francois Gaultier for the period 1742–1756 (Slonosky, 2003), although sporadic weather and temperature observations for the northern regions around Hudson's Bay exist before this date (e.g. Middleton, 1735). The next long-term, consistent weather record discovered so far is from Quebec City and starts in the 1790s (Spark, 1819; Lambert, 1984). The earliest long-term register for Montreal started in 1813 (McCord & McCord, 1826). With the exception of the years 1827 and 1828, a series of registers were kept in Montreal until the establishment of the professional Meteorological Service of Canada (MSC) in 1871.

The observations of 12 individuals, spanning 130 years, are considered in this article: Jean-Francois Gaultier (1742–1748; 1754), Alexander Spark (1798–1819), John Liveright (1823–1833), Thomas McCord (1813–1826), John Samuel McCord (1831–1842), Robert Cleghorn (1829–1833), William Skakel (1842–1852; 1862–1868), John Bethune (1838–1869), Louis-Edouard Glackmeyer (1844–1859), Archibald Hall (1863–1866), William Sutherland (1844–1848), and Charles Smallwood (1853–1862; 1868–1873).

The chief difficulty in comparing historical meteorological and climatological observations to modern data is in attempting to establish whether the historical and modern observations are both measuring the same thing: the state of the atmosphere at a given place. For the purposes of this article, ‘historical’ data will refer to the individual, largely amateur records kept in the 18th and 19th centuries before the establishment of the professional MSC, while ‘modern’ data will refer to the data collected by professional meteorologists after the foundation of the MSC in 1871. Changes in observing practice, instruments, standards of exposure of the instruments to the atmosphere and changes in the local environment often introduce bias into the measurements, such that the effect of these non-climatological measurement biases could obscure, exaggerate or distort any climatic signals. The problem is compounded when examining daily data as the high degree of temporal correlation in meteorological observations introduces time dependency in the individual observations. As most standard statistical methods for establishing differences in data sets require the data to be independently and identically distributed as a fundamental assumption, many standard statistical methods cannot be applied to daily observations. Instead, in this article, frequency histograms will be the chief method of examining daily observations and comparing historical data to modern data.

A discussion of 18th and 19th instruments and exposures is given in section 'Description of historical observations'. Descriptions of the weather registers of the historical observers are given in section 'Comparisons between historical and modern observations'. Section 'Data set quality and use' provides comparisons of the historical temperatures observations to modern data, and gives quality assessments of the historical observations.

This article describes in detail the instrumental observations and the procedures used to assess their quality. In a companion article, the temperature observations from the individual series described here are used to estimate minimum and maximum temperatures (Slonosky, 2014). These historical minimum and maximum temperature observations and estimates are then combined with homogenized data from Vincent et al. (2002) for the period 1873–2010 to form a single-temperature series from 1742 to 2010 (continuous from 1798 to 2010) for the St. Lawrence Valley region. This series is analysed for changes over time in both mean and extreme values in Slonosky (2014).

1 Thermometers and exposures

  1. Top of page
  2. Abstract
  3. Dataset
  4. Introduction
  5. 1 Thermometers and exposures
  6. 2 Description of historical observations
  7. 3 Comparisons between historical and modern observations
  8. 4 Data set quality and use
  9. Acknowledgements
  10. References
  11. Supporting Information

1.1 Instruments

The earliest observations in southern Canada were recorded by Gaultier at a time when thermometers as instruments were still experimental and in the process of being developed and refined. Gaultier's observations in Quebec helped to spur the development of an expanded scale, as the cold temperatures he recorded during his first winters at Quebec City were below the scale of his thermometer. His first thermometers were mercury-based and graduated in a calibration system devised by Delisle and used by l'Observatoire Royale de Paris. Thermometer scales at l'Observatoire de Paris in France were accordingly adjusted and graduated to capture the colder extremes of Canadian winters, and Gaultier's subsequent thermometers had expanded scales to better capture cold extremes. By 1754, Gaultier was using a spirit [alcohol]-based thermometer, described by Gaultier as being constructed according to the method of Réaumur, graduated with 54° above and below the freezing point of water (Gaultier, 1755). The use of Réaumur spirit-based thermometer had the added advantage of being able to record temperatures below −38 °C, the freezing point of mercury.

By the beginning of the 19th century, the mercury thermometer graduated in Fahrenheit degrees appears to be most commonly used in the Canadian records, although some Hudson Bay Company (HBC) posts were also provided with spirit thermometers (Wilson, 1982). It is likely that Liveright at the HBC post of Fort Coulonge also had a spirit thermometer as he recorded temperatures below the freezing point of mercury on several occasions. In the 1830s, McCord obtained his instruments directly from British instrument makers such as John Newman and Adie and Sons, and discussed issues of calibration with the instrument makers. McCord experimented with various different types of thermometers, including a spirit thermometer for measuring minimum temperatures. In John Samuel McCord's notebooks, a draft copy of a letter to the London instrument maker Newman discusses the behaviour of thermometers at low temperatures and attempts to determine the accuracy of different types of thermometers (spirit and mercury), suggesting that by the 1830s, calibration and accuracy, especially at extreme temperatures, was an issue of increasing importance (McCord, 1843). Errors increased with increasing cold, and were on the order of 3 °F too cold for horizontal mercury thermometers at 0 °F, but only about 0.5 °F for spirit (alcohol) thermometers (McCord, 1843). Similar calibration exercises were undertaken around this time by the Royal Society (Wilson, 1988).

Mercury thermometers were considered more reliable in general than spirit thermometers, so the Rutherford minimum–maximum thermometers were a combination of spirit minimum and mercury maximum thermometers (Middleton, 1966). Glackmeyer noted using a ‘self-registering’ Rutherford thermometer (Glackmeyer, 1859; Middleton, 1966). He recorded one of the coldest temperatures in the 19th century, −40 °F on 11 January 1859. Smallwood's thermometers are listed as being of ‘Sixes, Rutherford, Negretti, etc., the readings of which are corrected with the standard instruments of the new observatory, and most of the scales engraved on the stem of the tubes’ in the 1850s (Smallwood, 1858a). Smallwood recorded a temperature of −43.6 °F on 10 January 1859.

The frequency distributions of the original observations recorded in the weather registers (see Figures 1 and 2) show that until about 1830, there is a strong bias towards even numbers, suggesting thermometers were graduated at best to the nearest even number.

image

Figure 1. Frequency histograms of the original morning temperature observations for (a) Spark 8 a.m. (1300 UTC) (Quebec City) 1798–1819; (b) McCord 8 a.m. (1300 UTC) (La Grange aux Pauvres) 1813–1826; (c) Bethune minimum temperature (Montreal) 1838–1869; (d) Skakel 7 a.m. (1200 UTC) (Montreal) 1842–1852; (e) Glackmeyer sunrise (Quebec City region) 1844–1850; (f) Hall 7 a.m. (1200 UTC) (Montreal) 1863–1866.

Download figure to PowerPoint

image

Figure 2. As in Figure 1, for afternoon temperatures; (a) Spark 2 p.m. (1900 UTC); (b) McCord 4 p.m. (2000 UTC); (c) Bethune maximum temperature; (d) Skakel 3 p.m. (2100 UTC); (e) Glackmeyer noon; (f) Hall 2 p.m (1900 UTC).

Download figure to PowerPoint

1.2 Exposures

Until the mid-19th century the standard exposure of the thermometer adhered the principles laid down by Réaumur in his 1732 directions on how to make temperature observations from different places comparable (Réaumur, 1732). ‘It is absolutely essential’, Réaumur wrote, ‘that the Observer's thermometer must be exposed to the exterior air. The exposition chosen must be that of the North, and such that the sun cannot shine on it at any hour of the day. This alone is still insufficient unless the observer takes care that no nearby walls reflect sunlight onto the thermometer, and takes note as to whether the thermometer is placed on a second or third floor’. How closely the individual observers discussed below were able to follow these guidelines remains unknown in many cases, but notes in Gaultier's letters and Spark's and McCord's weather diaries show that Réaumur's principles were well known.

Frequency analysis and comparison to modern hourly observations suggest that in most cases the thermometers were adequately shielded from solar radiation. In one case, as demonstrated in section 2.8, Glackmeyer recorded that his thermometer was exposed to the afternoon sun, and this is clearly reflected in an increased frequency of high temperatures (see Figures 2(e) and 6(e)).

By the mid-19th century thermometer stands and shelters were coming into use. Charles Smallwood built an observatory on his property in the country village of St. Martin, 15 km north of the city of Montreal, where the thermometers were exposed to the air on the northern side of the observatory building and shielded both from above and on the sides by wooden louvered blinds (Smallwood, 1858a). Smallwood's instruments and observatory were transferred to McGill University in 1863, when the observatory was re-named the McGill Observatory and later the Montreal Observatory, and was one of the founding stations of the MSC in 1871.

1.3 North wall exposures

Many efforts have been made to determine the bias due to the typical north wall exposures prior to the 1850s and the various shelters and screen in use from the mid-19th century onwards (Gaster, 1879; Wilson, 1983; Chenoweth, 1993; Böhm et al., 2010; Brunet et al., 2010). To determine possible discrepancies between temperature readings taken from north wall exposures and those from sheltered open-space or garden exposures more typical of the late 19th and 20th centuries near Montreal, a series of modern parallel temperature observations with approximate reconstructions of these two types of exposures were maintained at the author's house for the years 2012–2013. Photographs of the exposures are shown in the Figure S8.

A small garden setting for a sheltered thermometer and an instrument exposed to an uninsulated north-facing brick wall may be reasonable replicas of historical urban settings. Figure 3 shows two examples of the largest differences found between north wall and screened thermometers, Figure 3(a) for summer (6–7 September 2012) and Figure 3(b) for winter (17–18 January 2013).

image

Figure 3. Temperature traces at one minute intervals showing the temperature of the thermometer with a north wall exposure (solid line) and the screened thermometer in an open-air setting (dashed line). (a) Temperatures from 0500 UTC 6 September 2012 to 1800 UTC 7 September 2012; (b) temperatures from 0500 UTC 17 January 2013 to 1800 UTC 18 January 2013.

Download figure to PowerPoint

In summer, the night-time differences are small, with the north wall exposure slightly warmer than the screened garden setting (Figure 3(a)). During the day, heat fluxes in the garden setting, such as heat emanating from the ground due to solar heating, raised the daytime temperature in an open-air setting above that of the north wall setting, which is well shaded during the day. This suggests that the shaded north wall exposure reduced the warming effect of radiative heat fluxes from the ground that were strongest in the exposed garden setting in calm and sunny conditions. Figure 3(b) shows that in winter the more sheltered north wall exposure can be up to 2 °C warmer than the screened thermometer on cold nights, a difference due to a combination of residual heat radiating from the house warming the north wall thermometer and strong radiative cooling affecting the screened thermometer. An unscreened thermometer (not shown) read −19.5 °C at 4 a.m. (0900 UTC) on 18 January 2003, compared to −18.8 °C for the screened thermometer and −16.8 °C for the north wall thermometer. On 18 January 2003, the north wall thermometer remained warmer than the screened thermometer until nearly noon, when short wave radiation warmed the open-air garden exposure to a higher temperature than the shaded north wall setting.

The temperatures from the north wall and screened exposures, recorded at one-minute intervals, were divided into four cases: day and night, determined by hours of local sunrise and sunset, and heating (15 October to 31 December 2012; winter) and non-heating (27 July to 14 October 2012; summer) seasons. Linear regression was performed on each of the four cases (winter night, winter day, summer night, summer day) using the north wall temperatures as a predictor and the screen temperatures as the predictand. The results are shown in Figure 4 and the regression values for each case are given in Table 1. These values were used to adjust the historical observations and were applied to those observations where either the thermometer was explicitly described as having a north wall exposure (Gaultier, Spark and Smallwood 1852–1863), or where the frequency distributions of minimum and maximum temperatures suggested the observations had a north wall bias (Skakel, Cleghorn and Liveright).

Table 1. Equations for north wall bias adjustments.
North wall adjustmentsEquation
  1. adj, adjustment; obs, observation.

Equation S1: daytime winter north wall adjustmentTadj = −0.89 + 1.027·Tobs
Equation S2: night-time winter north wall adjustmentTadj = −1.163 + 1.037·Tobs
Equation S3: daytime summer north wall adjustmentTadj = −1.47 + 1.064·Tobs
Equation S4: night-time summer north wall adjustmentTadj = −1.418 + 1.042·Tobs
image

Figure 4. Scatter plots showing the relationship between temperatures measured at the north wall exposure (x-axis) and with an open-air screened exposure (y-axis) between 27 July 2013 and 31 January 2013. Regression equations (Table 1) developed for (a) winter day; (b) winter night; (c) summer day; (d) summer night.

Download figure to PowerPoint

Similar parallel experiments with different thermometer exposures were performed by Böhm et al. (2010) in Austria and Brunet et al. (2010) in Spain. Brunet et al. (2010) found that in Spain, the open shelter of the Montsouris type was biased warm for maximum temperatures compared to a more sheltered Stevenson screen, with the greatest effect in summer. A shaded north wall setting is even more sheltered from the effects of daytime radiation fluxes than a Stevenson screen in an open environment, and the results presented above are consistent with those found by Brunet et al. (2010). Böhm et al. (2010) found the warm bias of north wall exposures as reconstructed in Austria to be greatest in the summertime morning and evening temperatures as the thermometers in their historical settings were inadqautely shielded from the rising and setting sun towards the northern horizon in summer.

In southern Quebec, the effect of snow cover leads to greater short and longwave radiation loss from the garden setting during winter, and to a greater difference (warm bias) in north wall compared garden exposures in winter, particularly for minimum temperatures. Average winter difference are 1.2°C (night) and 0.9°C (day) compared to 0.7°C (night) and 0.3°C (day) for summer. This contrasts with the findings of Böhm et al. (2010) and Brunet et al. (2010) who found the greatest difference between the various exposures to be in summer. The results found for Montreal generally agree with those of Gaster (1879) and with the discussion of north wall and screen biases in Wilson (1982), but continued observations are necessary to investigate this issue.

1.4 Quality control

Several methods were used to identify possible mistakes in the records, either in the documents used as primary sources or in the transcribing process. With a few exceptions (such as the Spark Diary and the McCord Family record), the majority of the sources are unlikely to be the original weather registers, but are more probably copies made by the observers or by other scientists such as J. S. McCord, who were interested in using these observations for climatic analysis. From notes recorded in the Spark diary and McCord's records, these observers had several instruments which they used in determining the temperature or pressure. Each step, therefore, allows the possibility of a transcription error to occur.

Values were compared against the previous observation, the same hour observation of the previous day, and to the modern average observation for the day. Outliers thus identified were checked against the original records. Monthly mean standard deviation values were also used to identify outliers or typographical errors.

Despite known discontinuities in many of the series, as for example when the observers recorded moving house or replacing broken thermometers, very few potential inhomogeneities are discernible either from visual inspection of time series or from difference series when two or more registers overlap. The lack of sufficient contemporary observations and the high variability of daily temperatures in this region make detecting potential inhomogeneities using difference series ineligible for many of the observations. Most known potential discontinuities are listed as address changes in Table 2, but are below the noise level of the observations and are not adjusted for in the historical series, with the exception of Spark (section 'Skakel 1842–1852, 1862–1869: city of Montreal'; Table 3).

Table 2. Observers, Locations and instruments. Description of each of the individual records used in constructing the 1742–2010 St. Lawrence Valley daily minimum and maximum temperature series. Elevations are estimated using Google Earth.
ObserverDatesLocationEnvironmentElevation (m)Variables observedInstruments (Estimated precision)Exposure
Gaultier1742–1748Québec Cityurban~657 a.m. and 3 p.m. temperature; wind direction; weathervarious; mercury (1° Delisle scale)outside window on north-facing wall
1754Réamur thermometer; alcohol (1° Réaumur (scale)
Spark1798–1801Québec CityUrban328 a.m. and 2–3 p.m. temperature and pressure (from 1803); wind direction; weatherUnknown thermometer; 2 barometers. (2 °F)Between double windows
1801–180732Outside window
1807–181957
McCord (Thomas and John Samuel) (McCord 1)1813–1826Montreal (outskirts)Rural; increasingly urbanized after 182410~8 a.m. and ~4 p.m. temperature; wind direction; weatherUnknown thermometer (2 °F)Unknown
Liveright1823–1833Fort Coulogne (Ottawa River)Rural107Sunrise, noon, and sunset temperature; wind direction; weatherUnknown thermometer (likely Six's) (2 °F)Unknown
Cleghorn1829–1833Montreal (outskirts)Rural43Sunrise, noon, and sunset temperature; weatherUnknown thermometer (2 °F) 
McCord (John Samuel) (McCord 2)1831–1832Montreal (city and outskirtsUrban~188 a.m., 5 p.m. temperature and pressure; wind direction; weatherVarious; Adie & Son standard, Newman's standard, vertical and horizontal register, spirit and min/max thermometers (0.1–1 °F)Instrument room (barometer); outdoor exposure (thermometer)
1832–1834Urban168 a.m., 5 p.m. temperature and pressure; wind direction and force; precipitation; weather
1835–18398 a.m., 9 p.m. temperature and pressure; min, max temperature; wind direction and force; precipitation; weather
1839–1842Urban20
July–September 1838, 1839Rural156
Bethune1838–1842MontrealUrban58Minimum and maximum temperature; pressure; wind; weatherUnknown, presumably max/min thermometer. (1 °F)Unknown
1842–1846Urban34
1846–1853Urban29
1853–1859Urban32
1859–1868Urban56
Skakel 11842–1852Montreal (city)Urban167 a.m., 3 p.m., 10 p.m. temperature and pressure; wind; weather(1 °F)Unknown
Skakel 21862–1869Urban
Sunderland1844–1848Montreal (city)Urban167 a.m., 3 p.m. temperature; wind; weather Unknown
Glackmeyer1844–1846BeauportRural96 Rutherford's day and night self-registering thermometer (1 °F)Thermometer exposed to sun at noon.
1846–1850Lower Town Quebec CityUrban56 a.m./sunrise, noon, 6 p.m./sunset temperature; wind; weather 
1850–1851CharlesbourgSuburban~125Thermometer exposed to setting sun
1851–1859CharlesbourgRural/Suburban~57 
Smallwood1849–1863St. Martin, Ile Jesus (Laval)Rural306 a.m., 2 p.m., 9 p.m. temperature and pressure; precipitation; wind; weather; cloud fraction, cloud type(1 °F)Observatory north wall shelter
Smallwood21868–1873*Montreal (McGill Observatory)Suburban587 a.m.,2 p.m.,9 p.m. temperature and pressure; precipitation; wind; weather; cloud fraction; cloud type(0.1 °F)Outside shelter
Hall1863–1866Montreal (city)Urban227 a.m., 2 p.m., 9 p.m. temperature and pressure; min max temperature; weather; wind(0.1 °F)Unknown
McGill Observatory1873–1963McGill ObservatoryUrban park58Min max temperature, precipitation(0.1 °F)Outside shelter
1964–1993Front lawn of McGill University; Physics building, Burnside HallUrban67
McTavish AWS1994–presentMcTavish Reservoir/Rutherford ParkUrban84Automatic weather station: hourly, min and max temperature, precipitation(0.1 °C)Automatic weather station; aspirated Stevenson screen.
Dorval (Pierre Elliot Trudeau) Airport1941–present (hourly 1953–present)City of Dorval, Montreal IslandUrban30Hourly, min and max temperature, precipitation, weather, wind, others(0.1 °C)MSC standard
Table 3. Data adjustments.
ObserverPeriodVariableAdjustment value
Spark19 June 1807–March 1819T8 a.m.+1.4 °C
Smallwood1852–1863T6 a.m.+1.0 °C
Smallwood 1868–18731868–1873T2 p.m., T9 p.m.−0.5 °F

Finally, histograms of the observations, converted to modern units, were constructed and compared to histograms of similar observations for modern data. Examples are given in Figures 5 and 6. These comparisons can help identify potential problems such as under-exposure to night-time radiative cooling effects or exposure to sunlight. Frequency distributions are also used to assess whether the data from the past and the data from the present are both measuring the same phenomena. While exact matches between the historical and modern distributions are not expected given the changing climate, the shorter records of the 18th and 19th century records, their more episodic nature, especially for the afternoon readings, and the differences in exposure and instrumentation, a comparison of the shape of the distribution can give an estimate as to whether the various historical registers are likely to be recording similar phenomena to the modern observations.

image

Figure 5. As in Figure 1, but for relative frequencies of morning temperature values converted to degrees celsius. (a) Spark 8 a.m. (Quebec City) 1798–1819; (b) McCord1 8 a.m. (La Grange aux Pauvres) 1813–1826; (c) Bethune minimum temperature (Montreal) 1838–1869; (d) Skakel 7 a.m. (Montreal) 1842–1852; (e) Glackmeyer sunrise (Quebec City region) 1844–1850; (f) Hall 7 a.m. (Montreal) 1863–1866. Dashed lines show the frequency modern values observed at the same time of day as the historical observers for Quebec City airport, 1953–2008 (Figure 2(a,e)) and Montreal (Dorval) airport, 1953–2008 (Figure 2(b,c,d,f)).

Download figure to PowerPoint

image

Figure 6. As in Figure 5, for afternoon temperatures; (a) Spark 2 p.m.; (b) McCord 1 4 p.m.; (c) Bethune maximum temperature; (d) Skakel 3 p.m.; (e) Glackmeyer noon; (f) Hall 2 p.m.

Download figure to PowerPoint

Observing times were not standardized until well into the 20th century, so each observer's set of readings (bars) is compared to the 1953–2008 frequency distribution of the comparable hourly observation (dashed line) in Figures 5 and 6. For the records which were observed at sunrise and sunset, the temperature at the closest hour to sunrise or sunset for the given day of the year was used.

2 Description of historical observations

  1. Top of page
  2. Abstract
  3. Dataset
  4. Introduction
  5. 1 Thermometers and exposures
  6. 2 Description of historical observations
  7. 3 Comparisons between historical and modern observations
  8. 4 Data set quality and use
  9. Acknowledgements
  10. References
  11. Supporting Information

The individual observers, their locations and the dates over which they recorded weather observations are listed in Table 2. The St. Lawrence Valley region is shown in Figure S1, with more detailed maps of the Montreal region and Quebec City region shown in Figures S2 and S3. Frequency distributions for observers not shown in Figures 1, 2, 5 and 6 can be found in Figures S4–S7.

2.1 Gaultier 1742–1748; 1754: Quebec City

The earliest set of observations, those of Jean-Francois Gaultier, has been described in detail in Slonosky (2003). Gaultier's daily weather and temperature observations can be found in papers published by l'Académie Royale des Sciences and in manuscript letters preserved in the archives of l'Observatoire de Paris and the Houghton Library of Harvard University (duHamel duMonceau, 1744, 1745, 1746, 1747; Gaultier, 1748, 1754). They consist of twice daily readings of morning and afternoon temperatures, although the morning observations are more consistent than the afternoon ones. Not only are 53% of potential afternoon observations missing, compared to 1.3% of morning observations, but those which were recorded in the afternoon often tended to be for some reason unusual: extremely cold, extremely warm, or indicating the arrival of the spring thaws.

Gaultier placed his thermometer according to Réaumur's instructions, outside a window on a north-facing wall of a cold room in which there was never a fire lit. As physician to the colony appointed by the Crown, his places of work were at the Citadel and l'Hôtel-Dieu (the hospital), and his residence was very likely to have been near these places in the Upper Town of Quebec City. The most likely location for Gaultier's observations is l'Hôtel-Dieu. There is some uncertainty as to the calibration of the thermometer used in 1754, and the current best estimate of calibration, based on concordance with documentary evidence, is that 1 degree of Gaultier's thermometer is about 9/10 of a degree Celsius.

2.2 Spark 1798–1819: Quebec City

The next series of continuous observations from a known source are those of Alexander Spark in his weather diary from 1798 to 1819 (Spark, 1819). A microfilm copy of this diary is kept at the McGill University Archives (MUA), and a second, manuscript partial copy is preserved at the McCord Museum (McM; McCord, 1836).

Spark recorded the temperature and weather twice daily at 8 a.m. and between 2 and 3 p.m. from 1798 to 7 March 1819, the day of his death. The Eastern Time Zone of Canada, where both Quebec City and Montreal are located, is 5 h behind UTC (UTC –0500). However, the measurement of time was not standardized until the late 19th century, so times should be considered as local and approximate relative to UTC. The type of instruments he used remains unknown, but he had more than one barometer and possibly more than one thermometer as well. His thermometer was at one point placed between double windows in his bedroom, but this was found to be too sheltered in January 1801, when he presumably moved the thermometer to outside his window (Spark, 1819). Spark's two known residences were in the Upper Town of Quebec City, and the house to which he moved in the spring of 1807 was within 300 m of his previous residence and of l'Hôtel-Dieu. An inhomogeneity was found to have ensued from this 1807 move to a site at a higher elevation, and 1.4 °C was added to the morning values from May 1807 until 1819 (see Table 3).

As with Gaultier, Spark was a professional man, a clergyman with duties which often called him from home during the day, and so his records are more complete for the 8 a.m. observations (3% missing) than the 2 p.m. ones (23% missing). For some periods, such as the summer of 1804, there are nearly no afternoon observations at all.

The frequency distributions of Spark's readings (Figures 1(a) and 2(a)) show a bias in favour of even numbers of nearly 4 : 1, strongly suggesting a thermometer graduated in even numbers only.

2.3 Thomas McCord 1813–1826, John Samuel McCord 1831–1842: Island of Montreal

The earliest continuous, long-term observations for Montreal are those recorded by the McCord family, starting in 1813 and continuing, with some gaps towards the end, until July 1826. It is not clear whether the observations in this first McCord register were kept by Thomas McCord or his son, John Samuel, or both, and will be referred hereafter as the McCord family record. The McCord family residence at this time was in a rural zone to the southwest of the contemporary city walls. The readings of the thermometer are often referred to as ‘the mercury’ in the manuscript (McCord & McCord, 1826). The times of the observations are noted, and are usually at 8 a.m. and 4 p.m., although they vary between 6 and 9 a.m. in the morning and 2 and 6 p.m. in the afternoon. Missing observations are again more common in the afternoon, with 22% of potential observations missing, compared to 5% for the morning values. Several thermometers were used over the course of 1813–1826, as thermometers were broken or, as on 1 June 1819, stolen. On exceptionally warm days, a second thermometer was placed in the sun, and a note was made on the temperature in shade and in the sun, and the time and reading of the maximum temperature noted. Before being sent to boarding school in Quebec City in 1816, where he was taught by a colleague of Alexander Spark, John Samuel McCord attended a day school run by Alexander Skakel (see section 2.7), and it is likely his observations methods were influenced by Skakel.

The strong bias towards even numbers shown in Figures 1(b) and 2(b), similar that seen in Spark's records, shows the graduation of the thermometers must have at the most every 2 °F, again giving a starting point for estimating the accuracy of the observations at about 1.25 °C. A thermometer in the McM collection is calibrated in increments of 2 °F on one side of the mercury tube, and 1°R on the other.

A second weather register kept by John Samuel McCord began in April 1831, continuing until June 1842. By this time J. S. McCord, a practicing lawyer and later a judge, was living in the heart of the old city of Montreal. In 1838 he built a country house, Temple Grove, on the upper slopes of Mont Royal, where he spent his summers. His register indicates that the weather observations for the summers of 1838 and 1839 were taken at Temple Grove (McCord, 1835, 1841, 1842, 1848).

Although it is known that McCord moved several times during the 1831–1842 period, and that during the summers of 1838, 1839, and probably 1840 and 1841 he spent the months of July–September on his country estate on the southern slope of Mount Royal ~3 km from his town house, it wasn't possible to discern a clear influence of this change in location on the temperature readings.

McCord's 1831–1842 observations are the first readings calibrated in Fahrenheit to show no bias towards even numbers, suggesting his thermometers were graduated in such a way as to be able to be read to the nearest 1 °F (not shown; see Figures S4(c) and S5(c)). His rate of missing observations is relatively high, with 7% of morning observation and 9% of afternoon observations missing.

2.4 Cleghorn 1829–1833: Island of Montreal

Robert Cleghorn kept a record of temperature and weather at his professional plant nursery, Blink Bonny Gardens, to the north-east of the contemporary city, a copy of which was made by J. S. McCord for the years 1829–1833 (Cleghorn, 1833). Cleghorn's observations were recorded at sunrise, noon, and sunset. The bias towards even numbers (not shown; see Figures S4(b) and S5(b)) is even more pronounced than with Spark and the first McCord record, with ratios of even to odd numbers approaching 5 : 1. As his instruments were at his place of work, however, and closely related to his botanical enterprise, his rate of missing observations is extremely low, with 0.7% of sunrise and 0.5% of noon readings missing.

Cleghorn's gardens were located to the north-east of the city, on a southern slope of Mount Royal, the hill in the middle of Montreal Island. Most of the observations discussed in this article were taken within an area of 1.5 km2, between the St. Lawrence River to the south-east and the southern slope of Mount Royal to the north-west. Cleghorn's gardens were located on the eastern side of a road leading across Mount Royal, with McGill University (the site of Montreal observations from 1868 onwards) on the western side of the road.

2.5 Liveright 1823–1833: Ft Coulonge (Ottawa River)

Daily weather observations consisting of temperature, wind and weather for 1823–1833 for Fort Coulonge, a HBC fur trading post on the Ottawa River 250 km WNW of Montreal, were also among J. S. McCord's scientific papers (Liveright, 1833). The HBC, advised by the Royal Society, instituted a programme of weather observations at its posts in 1814, in the hopes of making them self-supporting by growing food. Research by Wilson (1983, 1988) indicates the instruments favoured by the HBC in the early 19th century were either of the self-registering kind originally designed by James Six (Middleton, 1966) or mercury thermometers.

Liveright's observations, like Cleghorn's, were taken at sunrise, noon, and sunset. His observations also have a bias towards even numbers (not shown; see Figures S4(d) and S5(d)). He has very few missing observations, perhaps because recoding the weather was a professional duty of HBC post managers (Wilson, 1982), with missing value rates of 0.2–0.4%.

2.6 Bethune 1838–1869: city of Montreal

The longest continuous set of 19th century observations for Montreal is that of John Bethune. Bethune was principal of McGill University from 1835 to 1846, and later rector and then dean of Christ Church Cathedral. A copy of his observations of temperature, wind, weather and pressure in McCord's scientific papers span the middle of the 19th century from 1838 to 1869 (Bethune, 1869). Bethune's record is unusual in that he recorded minimum and maximum temperatures, rather than temperature observations made at regular, fixed hours as most of his contemporaries did. His rate of missing observations is also low, at 1.5% for minimum temperatures and 1.6% for maximum temperatures.

Bethune's observations were found as part of McCord's scientific papers, and there is little information beyond the observations themselves. It is not known at this time what the instruments were nor where the observations were taken, although presumably the thermometers were of a self-registering type. Bethune's various addresses over the period can be traced through 19th century directories, but the observations don't show any sign of breakpoints corresponding to the various dates of his changes of address, so it is possible the observations were taken at Christ Church Cathedral or McGill University, or that the changes of location within an area of less than 0.5 km2 were not significant enough to provoke detectable inhomogeneities.

Figures 1(c) and 2(c) suggest there is still a slight bias towards even numbers, but less pronounced than in earlier observations. The self-registering maximum and minimum thermometers used, especially towards the beginning of the record, may not have been graduated to as fine a scale as those of the register thermometers used by McCord at the same period.

2.7 Skakel 1842–1852, 1862–1869: city of Montreal

Among the weather records of the MUA are unattributed registers for 1842–1853 and 1862–1869 (Skakel, 1869). Monthly mean data for 1826–1835 attributed to Alexander Skakel were published in the Edinburgh Philosophical Journal (Hall, 1836) and later republished in Dove's Handbook of Climatology (Dove, 1840; Schott, 1876). Schott listed the source of monthly mean data for Montreal as ‘Wm Kakel’. Newspaper articles found in McCord's weather journals listed the coldest day of each year ‘recorded in Mr. Skakel's Meteorological Tables’ from 1820 to 1859 (McCord, 1838), while a similar newspaper clipping from Smallwood's journals recapitulates the warmest July temperatures since 1800, using the records of ‘Latour, McCord, Skakel and others’ (Smallwood, 1873). From various publications and newspaper articles, it is known that brothers Alexander and William Skakel had been recording temperature in Montreal since at least 1816 (Hall, 1836). The registers are therefore attributed in this paper to William Skakel, as Alexander Skakel died in 1846, based on the concordance of the observations in the above un-attributed manuscripts and the values printed in the newspaper tables. From Hall's (1836) article we learn that Alexander Skakel had been in the habit of making daily observations for the ‘last fifteen or twenty years; and moreover, his thermometer, having never during that time been removed from the station it now occupies… the observations, the one at 7 a.m., when it may be assumed the temperature is at minimum, the other at 3 p.m., when it may be stated to be at maximum…’ (Hall, 1836). This suggests the Skakel brothers’ records started at the latest in 1816.

Figures 1(d) and 2(d) indicate the observations from Skakel's first manuscript, covering the period 1842–1852 are of good quality: there is no bias towards even numbers and the bi-modal distribution is clearly indicated. The missing observation rate is less than 2%.

The quality of observations deteriorated towards the end of the second Skakel manuscript (1862–1869), with values repeated for successive days or the second digit of a reading not recorded. It is highly likely that this manuscript is a copy of the original record, and it is possible to copyist found some numbers to be unreadable. It is not known who continued the record after William Skakel's death in 1868, although a note in the MUA suggests the register was kept for a time at the High School of Montreal, adjacent to McGill University and near the location of Cleghorn's observations. Frequency distributions for the second Skakel record (not shown; see Figures 4(e) and 5(e)) show a less well-distributed series, with fewer extreme values and a possible observation bias. The missing observation rate is 5%, which includes values which were illegible or improperly copied in the manuscript and removed from the record analysed here.

2.8 Glackmeyer 1844–1859: Quebec City and surroundings

A later long-term daily weather diary for the Quebec City region is that of Louis-Edouard Glackmeyer for 1844–1859 (Glackmeyer, 1859). A notary, Glackmeyer moved several times, starting his observations at Beauport, a town some 8 km from Quebec City and now considered a suburb of Quebec City. In 1846, Glackmeyer moved to the Lower Town of Quebec City, and in 1851 to Charlesbourg, another suburb about 7 km to the west of Quebec City. Glackmeyer noted that he had a Rutherford self-registering day and night thermometer, but his reported values are for fixed times. Glackmeyer's missing value rate is 1% for morning, 7% for noon, and 2% for evening observations.

His morning observations (Figure 1(d)) show a bias towards both even numbers and also towards multiples of 5, suggesting the major markings on his minimum thermometer may have been every 5 or 10 °F. The afternoon readings show a bias for numbers ending in multiples of 10 (Figure 2(e)).

2.9 Sutherland 1844–1848: city of Montreal

Dr. William Sutherland's series runs from 1844 to 1848 (Sutherland, 1848). While some observations come from manuscripts in the MUA, medical practitioners (Sutherland, Hall, and Smallwood especially) also published their meteorological observations at the end of the monthly issues of contemporary scientific journals, particularly the Canadian Journal and the British American Journal. Sutherland's missing values rate is around 4%. As Sutherland's is a short record, there is considerable noise and the characteristic shape of the double peak is less defined: there is a higher frequency of warm values (~24–25 °C) and cold temperatures (<−10 °C) compared to modern values (not shown; see Figures S4(f) and S5(f)).

2.10 Smallwood 1853–1862: Island of Laval (Ile-Jésus); 1868–1873 (McGill University, Montreal)

Dr. Charles Smallwood built a sophisticated observatory near his home on the island of Laval, situated in the St-Lawrence River to the north of the island of Montreal, described in detail in Smallwood (1858a) and Marshall and Bignell (1969). Smallwood's observations have been recovered for two periods: 1849–1862 at his observatory in St-Martin and 1868–1873 at the McGill College Observatory. The observations from 1849 to 1852 are only sporadically noted at a variety of different times, making them difficult to evaluate as a whole. The observations from 1853 to 1862 show a larger range than most of the observations recorded on the island of Montreal, which could be due either to local micro-climate issues, a more rural setting, or biases due to Smallwood's exposure (not shown; see Figures S4(g) and S5(g)). Temperature was measured three times daily and published, from 1853 to 1862, in various contemporary scientific journals, chiefly the Canadian Journal. There are virtually no missing observations for 1853–1862.

Smallwood's thermometers were hung on the north-facing wall of his observatory building, which was some 20 m away from his house. He had Venetian screens on either side of the north wall to shield the thermometers from summertime short-wave radiation. The observatory building was mostly unheated, although on occasion a small charcoal stove or spirit lamp was used in various electrical experiments or to melt snow to measure its water equivalent. The thermometers are thus likely to be biased cold for the night and early morning observations, as the exposed thermometer bulb would cool from long-wave radiation to the environment without the compensating effects of heat retention typical of the north wall exposure of an occupied building. In 1857, Smallwood wrote that the highest temperature he recorded of the past seven years at St. Martin was 100.1 °F (37.7 °C), and the lowest −36.2 °F (−37.9 °C) (Smallwood, 1858b).

Experiments by K. Devine (K. Devine, pers. comm) suggest that exposed thermometers read about 1 °C cooler at night and in the early morning than screened thermometers. Rather than applying the north wall correction for the morning temperatures, 1 °C was therefore added to Smallwood's 6 a.m. observations for 1852–1863 and the afternoon values were left unadjusted (see Table 3).

Smallwood was invited to become the first professor of Meteorology at McGill University in 1863, declining a professorship in the Faculty of Medicine to do so. He moved his observatory to the campus of McGill University, where it was incorporated into the national meteorological network and remained the principal observing station for Montreal until 1963. In 1871, with the establishment of the MSC, meteorological observations became a professional concern and were standardized and collected at a national level (see Vincent et al., 2002).

A handwritten copy of meteorological observations from the McGill Observatory for 1868–1873, with Smallwood's notes and commentaries, is available in the MUA, with fixed hour readings for 6 a.m., 2 p.m. and 9 p.m. (Smallwood, 1873). Copies of these observations, together with minimum and maximum temperatures, are also available in the MSC archives starting in 1871, although the minimum and maximum temperatures appear unreliable. For the purposes of this paper, the 1868–1873 values from the MUA are assumed to be the original readings. The missing observation rate is 3% for both morning and afternoon readings. The weather registers for the 1870s observations at McGill College Observatory held at the MUA show that the thermometers had a bias of 0.5 °F. Comparisons with contemporary observers suggest that while the morning readings for 1868–1873 were similar to Bethune and Skakel, the afternoon and evening readings were slightly too high; 0.5 °F was accordingly removed from Smallwood's afternoon and evening observations (see Table 3).

2.11 Hall 1863–1866: city of Montreal

Dr. A. J. Hall was also a medical weather observer. His observations from 1863 to 1866 can be found in the MUA (Hall, 1866). Hall was a regular contributor to the Smithsonian Institute's (SI) weather observation programme, and copies of his observations can also be found in the SI archives. He also published his and Skakel's observations in medical journals, often with tables of mortality attached (e.g. Hall, 1847). Nineteenth century medical science was extremely interested in the connection between weather and disease, and this interest led many doctors to keep meteorological observations. The extremely short record from 1863 to 1866 makes it difficult to evaluate the quality of the observations. The short record and its sporadic nature show a distribution which appears to be skewed towards the extremes, although no instrument or observing bias appears to be present (Figures 1(f) and 2(f)).

2.12 Modern data

Readings were continued at the McGill College Observatory, later renamed the Montreal Observatory, in January 1874 under the direction of the MSC, although only minimum and maximum temperatures are available in digital format at present.

Temperatures recorded every hour at Canadian airports are available from the MSC Canadian Daily Climate Data Online (CDCD) website from 1953 onwards (http://climate.weather.gc.ca). Hourly data for the modern period (1953–2008) from the Montreal International Airport at Dorval and from the Quebec City Airport are used to compare the historical observations with modern temperatures.

The maximum and minimum temperatures from Quebec City and the McGill Observatory in Montreal have undergone extensive quality control and homogenization by Vincent et al. (2002, 2012). The values used here are taken from the 2002 data set, with updates added from the CDCD website.

3 Comparisons between historical and modern observations

  1. Top of page
  2. Abstract
  3. Dataset
  4. Introduction
  5. 1 Thermometers and exposures
  6. 2 Description of historical observations
  7. 3 Comparisons between historical and modern observations
  8. 4 Data set quality and use
  9. Acknowledgements
  10. References
  11. Supporting Information

Figures 5 and 6 show the relative frequencies in percentages of the historical observations (bars) with modern observations from 1953 to 2008 for the same time of day as the historical readings (dashed lines) for the morning and afternoon observations. Six observers are shown, providing a sampling of observations from 1798 to 1869.

The frequency distributions for both Montreal and Quebec City show double peaks, one centred around 0 °C, the second between 15 °C and 20 °C. When the observations are separated into winter and summer seasons, each season shows a normal distribution (not shown); the double peak of the entire series is a result of the superimposition of a broad normal distribution for the winter season and narrower one for the summer season. This pattern is characteristic of the region. Bergström and Moberg (2002) suggest that in cold climates, the high frequency of observations around 0 °C might be related to periods of melting snow in spring where the phase transition of water can constrain temperatures to around the freezing point. Both Montreal and Quebec City are subject to freeze–thaw cycles during the winter season.

Both Spark and McCord (Figures 5(a,c) and 6(a,c)) show frequency distributions similar to the modern distribution, which is especially remarkable given the uncertainty in the instrumentation, exposure, and time of observation. This increases our confidence that these observers and the modern observations are indeed both measuring similar underlying weather and climatic events, and the historical observations are not unduly affected by issues of instrumentation or exposures. The morning values for Spark, and to a lesser degree, McCord, show higher relative frequencies of temperatures below 0 °C, suggesting colder temperatures during this period. As both the McCord and Spark records contain the cold decade of the 1810s, this is perhaps not surprising. McCord's afternoon temperatures are also slightly colder than the 20th century values, but Spark's afternoon temperatures are slightly shifted to the right, suggesting a warm bias in winter, possibly from the north wall exposure. When the north wall adjustments are applied, the winter values, especially towards the extremes, are closer to the modern distribution (Figures 5(b) and 6(b)). This north wall bias is not seen in the McCord record.

With over 30 years of observations, the relative frequency distribution of Bethune's values are smoother than for records of shorter duration, and coincide closely with the 20th century distribution (Figures 5(c) and 6(c)). There is a slightly higher incidence of extremely cold minimum temperatures seen in Bethune's values but no evidence of a north wall bias or other instrumental problems.

The values shown for Glackmeyer are for the years 1844–1849, as it is known he moved in 1846 and again in 1850. His values show a slight warm bias in the morning observations (Figure 5(e)), but an extreme shift to the right for the noon temperatures (Figure 6(e)). In his manuscript, the thermometer is recorded as being exposed to the sun at noon but sheltered from radiation in the morning and the evening from 1844 to 1846. During 1850–1851 the thermometer was exposed to the setting sun. Not only is the entire distribution moved towards higher values, but the characteristic double peak is missing in the noon observations, and is less pronounced than with the other observers for the sunrise observations. The frequency distributions of the Glackmeyer record for the individual locations listed in Table 2 (not shown) and for the entire record don't fit with the overall distribution of temperatures either for the modern period or the various other periods of historical data for the Quebec City region. Correlations (not shown) also do not show as much coherence with contemporary observations as do most of the other records. Although the Glackmeyer record contains much other valuable information, and the morning observations may not be as badly affected by solar radiation, the record is not considered further for temperature analysis.

Hall's observations (Figures 5(f) and 6(f)) also do not correspond to the typical shape of temperature distributions of the St. Lawrence Valley region, with a very large number of both warm and cold extremes. However, Hall's short record and missing observation rate of around 30% affect the shape of the distribution, making comparisons using relative frequencies difficult. Correlations between Hall's observations and those of contemporary observers such as Bethune or Skakel are high, suggesting that the individual observations may be reliable. Without further information on Hall's instruments or exposures, the ability to assess the overall quality of the series, due to its short and sporadic nature his observations, must be considered as somewhat uncertain.

4 Data set quality and use

  1. Top of page
  2. Abstract
  3. Dataset
  4. Introduction
  5. 1 Thermometers and exposures
  6. 2 Description of historical observations
  7. 3 Comparisons between historical and modern observations
  8. 4 Data set quality and use
  9. Acknowledgements
  10. References
  11. Supporting Information

The examples in section 3.2 are given to show the strengths and weakness of using frequencies analysis to assess the quality of historical observations. When the historical records are well sampled, this technique can be used to determine information regarding the quality and characteristics of the thermometers used. Examination of the distribution of the original observations as recorded in the manuscripts revealed that most of the thermometers before the 1830s were graduated only to the nearest 2 °F. This gives an indication of the precision of the observations in the absence of other information regarding the instruments used. A confirmation of this surmise, obtained from an examination of the statistical properties of the raw data, was found in a thermometer from this period conserved in the McM which is graduated in both degrees Reaumur and Fahrenheit, the Fahrenheit scale being graduated every 2 °F, with the major markings every 10 °F. The distribution of the data can also show whether the extremes of temperature are well represented in the original observations.

Further assessment of the quality of the data as well as likely instrument exposures and other details of the historical observations can be determined by comparing the relative frequency distribution of the historical data to that of modern day observations. It should be noted that it is necessary to have as close a match as possible between the hours of observation in the historical and modern data. Hourly observations for both Montreal and Quebec City are available since 1953, providing over 20 000 individual observations for each hour of the day with which to determine the frequency distribution of temperature for a given hour at these locations. By comparing the distributions of historical and modern temperatures recorded at the same hour, it was possible to detect biases due to instrument exposure. An example of a north wall exposure bias could be seen in an under-representation of cold temperatures in winter in Spark's afternoon values. An extreme example of exposure to solar radiation was seen in Glackmeyer's noon temperatures and was confirmed by evidence in the manuscript. These examples give confidence that this method can detect instrument or exposure biases in historical data, as well as increasing confidence in those observations where the historical and modern distributions are well matched.

On the other hand, this method depends on both the availability of hourly data in the modern period and on the assumption that the historical observations are well-sampled, with no sampling bias. This is not always the case, especially with observers who were not professional meteorologists and whose other duties sometimes interfered with their ability to keep records. Both Gaultier and Hall showed sampling biases, particularly in their afternoon readings, when readings were more likely to be recorded during unusual conditions of especial interest to the observer. Gaultier recorded afternoon temperatures more often when they were unusually warm or when the spring thaws set in. From the distribution of his observations, Hall appears to have recorded the temperature more often when it was either unusually warm or unusually cold. Short records also exacerbate this problem, as a given period of three or four years may in itself have been warmer or colder than normal. This sampling bias does not mean that the individual readings are problematic. Hall's readings, for example, in general correlate well with contemporary observations, while Gaultier's observations are often corroborated by descriptive information in his reports.

This analysis shows that most of the observations from the 18th and 19th century can be considered reliable indicators of morning and afternoon temperature in the St. Lawrence Valley region. The observations from Gaultier have a greater uncertainty due to the unknown scale, and those from before 1830 are less precise due to the graduation of the thermometers, but nevertheless reflect the same characteristics seen in later observations.

To make the observations comparable to one another, as well as to long-term temperature measurements for the 20th century, a common variable must be used. The different observation methods, particularly the many different observing schedules throughout the 18th and 19th centuries, together with the long series of measured minimum and maximum temperatures from Bethune (1838–1869) and the modern period (1873–present) make estimates of minimum and maximum temperatures the most suitable means of comparing the various sets of observations. However, because of the highly variable nature of temperature in this region of the world, and the frequent passage of weather systems often leading to atypical changes in temperature throughout the 24-h period, the estimation of minimum and maximum temperatures from fixed hour observations is not straightforward and is described in detail in a companion paper (Slonosky, 2014).

A single series of daily minimum and maximum temperature for the St. Lawrence Valley region compiled from these historical observations is also discussed in Slonosky (2014), and the variability and changes in temperature shown by the observations described here are explored further. The availability of daily data for over 200 years makes it possible to analyse the variability in temperature extremes as well as mean values. The sub-daily nature of the manuscript records allows the analysis of changes in both the minimum and the maximum temperatures over time. Daily data from the St. Lawrence Valley region can also be used in conjunction with information from other locations to study in detail specific weather events, such as the well-known cold summer of 1816, in a century-scale context.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Dataset
  4. Introduction
  5. 1 Thermometers and exposures
  6. 2 Description of historical observations
  7. 3 Comparisons between historical and modern observations
  8. 4 Data set quality and use
  9. Acknowledgements
  10. References
  11. Supporting Information

Thanks are due to Cynthia Wilson for her encouragement. Thanks are also due to Josée Alexandre of l'Observatoire de Paris, Gordon Burr and Theresa Rowat of McGill University, Stephanie Poisson and Nora Hague of the McCord Museum, Joan Self of the UK National Meteorological Archive, Morley Thomas, Lucie Vincent and Anna Deptuch-Staph of Environment Canada and Cary Mock of the University of South Carolina for help with historical sources. Bill Hogg, Francis Zwiers and Val Swail from Environment Canada and Elizabeth Piper of NiCHE supported data digitization. Bob Jones of CMOS and Gavin Schmidt helped recruit volunteers. Gilles Paquette, Ray Couture, Pat Fortin, Rose Dlhopolsky, Carolyn Verduzco, Jennifer Dowker, Kristin Davoli, Alana Cameron, Kyle Hipwell, Dan Manweiler and Nancy Hagen all generously volunteered their time to digitize data. Juerg Luterbacher and Athanasios Tsikerdekis of the University of Giessen also supported and worked on the data digitization. André Plante provided valuable statistical advice and Herve Hacot provided technical support. Thanks to Lucie Vincent, Phil Jones and two anonymous reviewers for comments on the text.

References

  1. Top of page
  2. Abstract
  3. Dataset
  4. Introduction
  5. 1 Thermometers and exposures
  6. 2 Description of historical observations
  7. 3 Comparisons between historical and modern observations
  8. 4 Data set quality and use
  9. Acknowledgements
  10. References
  11. Supporting Information
  • Bergström H, Moberg A. 2002. Daily air temperature and pressure series for Uppsala (1722–1998). Climatic Change 53: 213252, doi:10.1023/A:1014983229213.
  • Bethune J . 1869. Fonds McCord Family, P001-835, “John Bethune's Meteorological Reports”. McCord Museum, Montreal, QC, Canada.
  • Böhm R, Jones PD, Hiebl J, Frank D, Brunetti M, Maugeri M. 2010. The early instrumental warm-bias: a solution for long central European temperature series 1760–2007. Climatic Change 101: 4167, doi:10.1007/s10584-009-9649-4.
  • Brunet M, Asin J, Sigró J, Bañón M, García F, Aguilar E, Palenzuela JE, Peterson TC, Jones P. 2010. The minimization of the screen bias from ancient Western Mediterranean air temperature records: an exploratory statistical analysis. International Journal of Climatology 31: 18791895, doi:10.1002/joc.2192.
  • Chenoweth M. 1993. Nonstandard thermometer exposures at US cooperative weather stations during the late nineteenth century. Journal of Climate 6: 17871797.
  • Cleghorn R. 1833. Fonds McCord Family, P001-839, “Copies of Weather Diaries”. McCord Museum, Montreal, QC, Canada.
  • Dove HW. 1840. Ueber die nicht periodischen Änderungen der Temperaturvertheilung auf der Oberfläche der Erde, in dem Zeitraume von 1789 bis 1838. G.E. Reimer: Berlin, Germany.
  • Gaster F. 1879. Report on experiments with thermometer screens. Quarterly Weather Report Appendix I: 1–11.
  • Gaultier J-F. 1748. “Journal Des Observations Météorologiques de M. Gaultier à Kebec Depuis Le 1 Octobre 1747 Jusqu'au 1 Octobre 1748”. Fonds Joseph-Nicolas Delisle. Acc # 64-6-A. Observatoire de Paris, Paris.
  • Gaultier J-F. 1754. “Journal Des Observations Météorologiques de M. Gaultier à Kebec 1754”. Meteorological Collection. Acc # Doc. No. 57M-13. Houghton Library Harvard University, Boston, MA, USA.
  • Gaultier J-F. 1755. “Observations on Quebec, 1744-1754”. Meteorological Collection. Acc # MS Can 42. Houghton Library Harvard University, Boston, MA, USA.
  • Glackmeyer L-E. 1859. “Meteorological Observations Taken at Beauport Near Quebec”. Meteorology. Acc # R6707-0-5-E. Library and Archives Canada, Ottawa, ON, Canada.
  • Hall A. 1836. Mean temperature of Montreal, Lower Canada for the period of Ten years, viz, from 1826 to 1835 inclusive. Edinburgh New Philosophical Journal 1836: 236239.
  • Hall A. 1847. Meteorological observations. British American Medical Journal 1847: 27.
  • Hall A. 1866. “The Observations of Dr. Hall”. Faculty of Arts and Sciences, Department of Meteorology, 1798-1972, Meteorological Records, 1798-1972. Container 955, Envelope 324a. McGill University Archives, Montreal, QC, Canada.
  • duHamel duMonceau HL. 1744. Observations botanico-météorologiques faites à Québec par M. Gautier pendant l'année 1743. Mémoires de l'Academie Royale des Sciences 1744: 135155.
  • duHamel duMonceau HL. 1745. Observations botanico-météorologiques faites à Québec par M. Gaultier pendant l'année 1744. Mémoires de l'Academie Royale des Sciences 1745: 194229.
  • duHamel duMonceau HL. 1746. Observations botanico-météorologiques faites à Québec par M. Gaultier pendant l'année 1745. Mémoires de l'Academie Royale des Sciences 1746: 8897.
  • duHamel duMonceau HL. 1747. Observations botanico-météorologiques faites à Québec par M. Gaultier pendant l'année 1746. Mémoires de l'Academie Royale des Sciences 1747: 466488.
  • Lambert JH. 1984. One man's contribution: Alexander Spark and the establishment of Presbyterianism in Quebec 1784-1819. Quebec, Canada.
  • Liveright J. 1833. Fonds McCord Family, P001-838 “John Liveright's Thermometrical Journals”. McCord Museum: Montreal, QC, Canada.
  • Marshall JS, Bignell N. 1969. Dr Smallwood's weather observatory at St. Martin's. Naturaliste Canadien 96: 483490.
  • McCord JS. 1835. Fonds McCord Family, P001- 829, “Meteorological Tables”. McCord Museum, Montreal, QC, Canada.
  • McCord JS. 1836. Fonds McCord Family, P001-825 “Scientific Notebook”. McCord Museum, Montreal, QC, Canada.
  • McCord JS. 1838. “Letter to Dr Shey”. McCord Family Papers 1766-1945. Acc # 0827. McCord Museum, Montreal, QC, Canada.
  • McCord JS. 1841. Meteorological Summary of the Weather at Montreal, Province of Canada (from 1836-1840). American Journal of Science and Arts XLI: 330331.
  • McCord JS.1842. Fonds McCord Family, P001- 830 “Meteorological Tables”. McCord Museum, Montreal, QC, Canada.
  • McCord JS. 1843. Fonds McCord Family, P001- 828, “Notebook on Climate and Meteorology of North America”. McCord Museum, Montreal, QC, Canada.
  • McCord JS. 1848. Observations on meteorology. British American Medical and Physical Journal 1848: 228229.
  • McCord T, McCord JS. 1826. “McCord's Meteorological Register”. Faculty of Arts and Sciences, Department of Meteorology, 1798-1972, Meteorological Records, 1798-1972. Container 1039, Envelope 320. McGill University Archives, Montreal, QC, Canada.
  • Middleton C. 1735. Observations made of the latitude, variation of the magnetic needle, and weather, by Capt. Christopher Middleton, in a voyage from London to Hudson's-Bay, Anno 1735. Philosophical Transactions of the Royal Society 39: 270280.
  • Middleton WEK. 1966. A History of the Thermometer and its Use in Meteorology. Johns Hopkins Press: Baltimore, MD.
  • Réaumur RAF. 1732. Régles pour construire des thermomètres dont les degrés sont comparables. Mémoires de l'Academie Royale des Sciences 1732: 452507.
  • Schott CA. 1876. Tables, distribution and variations of the atmospheric temperatures in the US and adjacent parts of America. Smithsonian Institution: Washington, DC.
  • Skakel W. 1869. “Meteorological Observations”. Faculty of Arts and Sciences, Department of Meteorology, 1798-1972, Meteorological Records, 1798-1972. Container 1039, Envelopes 318-319. McGill University Archives, Montreal, QC, Canada.
  • Slonosky V. 2003. The meteorological observations of Jean-Francois Gaultier, Quebec, Canada: 1742-56. Journal of Climate 16: 22322247.
  • Slonosky VC. 2013. St. Lawrence Valley, Quebec 18th and 19th Century Daily Temperature Data. NOAA/NCDC: Boulder, CO. 5654. http://hurricane.ncdc.noaa.gov/pls/paleox/f?p=519:1:::::P1_STUDY_ID:15654 (accessed 12 December 2013).
  • Slonosky V. 2014. Daily minimum and maximum temperature in the St-Lawrence Valley, Quebec: Two centuries of climatic observations from Canada. International Journal of Climatology (under review).
  • Smallwood C. 1858a. The Observatory at St. Martin, Isle Jesus, Canada East. Canadian Naturalist and Geologist 1858: 352363.
  • Smallwood C. 1858b. The meteorology of the Vicinity of Montreal. Proceedings of the American Association for the Advancement of Science 1858, 197204.
  • Smallwood C. 1873. “Dr Smallwood's Observations 1868-1873”. Faculty of Arts and Sciences, Department of Meteorology, 1798-1972, Meteorological Records, 1798-1972. Container 1039, Envelopes 326-327. McGill University Archives, Montreal, QC, Canada.
  • Spark A. 1819. “Dr Spark's Meteorological Register”. Faculty of Arts and Sciences, Department of Meteorology, 1798-1972, Meteorological Records, 1798-1972. Container 1039, Envelopes 318-319. McGill University Archive, Montreal, QC, Canada.
  • Sutherland W. 1848. “The Observations of Dr. Sutherland”. Faculty of Arts and Sciences, Department of Meteorology, 1798-1972, Meteorological Records, 1798-1972. Container 1039, Envelope 321. McGill University Archives, Montreal, QC, Canada.
  • Vincent LA, Zhang X, Bonsal BR, Hogg WD. 2002. Homogenization of daily temperatures over Canada. Journal of Climate 15: 13221334, doi:10.1175/1520-0442.
  • Vincent LA, Wang XL, Milewska EJ, Wan H, Yang F, Swail V. 2012. A second generation of homogenized Canadian monthly surface air temperature for climate trend analysis. Journal of Geophysical Research 117:19842012, doi: 10.1029/2012JD017859.
  • Wilson C. 1982. The Summer Season Along the East Coast of Hudson Bay during the Nineteenth Century: A sample study of small-scale historical climatology based on the Hudson's Bay Company and Royal Society Archives. Canadian Climate Centre Report. Atmospheric Environment Service: Downsview, ON, Canada; 85-3, 223 pp.
  • Wilson C. 1983. Some aspects of the calibration of early Canadian temperature records in the Hudson's Bay Company archives: a case study for the summer season, eastern Hudson/James Bay, 1814 to 1821. In Syllogeus 49: Climatic Change in Canada 3: 144–201. Canadian Museum of Nature: Ottawa, ON, Canada.
  • Wilson C. 1988. The Summer Season Along the East Coast of Hudson Bay during the Nineteenth Century: Part III Summer thermal and wetness indices. B. The indices, 1800 to 1900. Canadian Climate Centre Report. Atmospheric Environment Service: Downsview, ON, Canada; 85-3, 41 pp.

Supporting Information

  1. Top of page
  2. Abstract
  3. Dataset
  4. Introduction
  5. 1 Thermometers and exposures
  6. 2 Description of historical observations
  7. 3 Comparisons between historical and modern observations
  8. 4 Data set quality and use
  9. Acknowledgements
  10. References
  11. Supporting Information
FilenameFormatSizeDescription
gdj311-sup-0001-FigS1-S8.docxWord document8834K

Figure S1. The St. Lawrence Valley region, with Fort Coulonge on the Ottawa River also shown. Figure taken from Google Earth 2013: Google Earth 7.1. St-Lawrence Valley 42°N, 75°W, Satellite layer.

Figure S2. Montreal and region. Figure taken from Google Earth 2013: Google Earth 7.1. Montreal, Qc 45°N, 73°W, Satellite layer.

Figure S3. Quebec City and region. Figure taken from Google Earth 2013: Google Earth 7.1. Quebec City, Qc 47°N, 72°W, Satellite layer.

Figure S4. Frequency histograms of the original morning observations for (a) Gaultier 8 a.m. (Quebec City) 1742–1748, 1754; (b) Cleghorn sunrise (Blink Bonny Gardens) 1829–1833; (c) McCord2 8 a.m. (St James Street) 1831–1842; (d) Sutherland 7 a.m. (Montreal) 1844–1848; (e) Liveright sunrise (Fort Coulogne) 1823–1833; (f) Glackmeyer sunrise (Quebec City region) 1844–1859.

Figure S5. Frequency histograms of the original afternoon observations for (a) Gaultier 2 p.m. (Quebec City) 1742–1748, 1754; (b) Cleghorn noon (Blink Bonny Gardens) 1829–1833; (c) McCord2 9 p.m. (St James Street) 1831–1842; (d) Sutherland 3 p.m. (Montreal) 1844–1848; (e) Liveright noon (Fort Coulogne) 1823–1833; (f) Glackmeyer noon (Quebec City region) 1844–1859.

Figure S6. Relative frequencies of morning temperature values converted to Celsius degrees (a) Gaultier 8 a.m. (Quebec City) 1742–1748, 1754; (b) Cleghorn sunrise (Blink Bonny Gardens) 1829–1833; (c) McCord2 8 a.m. (St James Street) 1831–1842; (d) Sutherland 7 a.m. (Montreal) 1844–1848; (e) Liveright sunrise (Fort Coulogne) 1823–1833; (f) Glackmeyer sunrise (Quebec City region) 1844–1859. Dashed lines show the frequency modern values observed at the same time of day as the historical observers for Quebec City airport, 1953–2008 (Figure 6(a,f)) and Montreal (Dorval) airport, 1953–2008 (Figure 6(b–e)).

Figure S7. Relative frequencies of afternoon values converted to Celsius degrees. (a) Gaultier 2 p.m. (Quebec City) 1742–1748, 1754; (b) Cleghorn noon (Blink Bonny Gardens) 1829–1833; (c) McCord2 9 p.m. (St James Street) 1831–1842; (d) Sutherland 3 p.m (Montreal) 1844–1848; (e) Liveright noon (Fort Coulogne) 1823–1833; (f) Glackmeyer noon (Quebec City region) 1844–1859. Dashed lines show the frequency modern values observed at the same time of day as the historical observers for Quebec City airport, 1953–2008 (Figure 6(a,f)) and Montreal (Dorval) airport, 1953–2008 (Figure 6(b-e)).

Figure S8. (a) position of the north wall thermometer located near a window next to an uninsulated brick wall; (b) Thermometer shelter in a garden setting. The screen is located approximately 8m to the north of the wall mounted thermometer.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.