Katsue Misawa's sunspot observations in 1921–1934: a primary reference for the Wolfer–Brunner transition

ABSTRACT

Our knowledge of the centennial solar variability is largely based on the time series of international sunspot number (⁠|$S_{N}$|⁠), a composite index based on multiple visual sunspot observers’ records from the 18th century onward and maintained by the World Data Center (WDC) ‘Sunspot Index and Long-term Solar Observations’ (SILSO). However, over the period 1919–1944, our capacity to diagnose the homogeneity of this time series is currently limited, because most of the archived source data of the Zürich Observatory were never published over that interval and are presently still missing. Therefore, it is essential to recover any long-duration series from individual sunspot observers active during this period to bridge this Zürich data gap. In this context, Katsue Misawa has conducted regular sunspot observations from 1921 to 1934 (mean coverage of 25.4 days per month), which were not accessible for the Zürich Observatory and thus form a valuable addition to the data base maintained by the WDC–SILSO. In this study, we digitized his observational records, documented his observing technique, and reconstructed his total and hemispheric SNs. We compared his data with the international S_N (current version V2) and evaluated their stability. Misawa's data series generally agrees well with S_N V2. However, Misawa's data show a significant transitory drift in 1925–1928 against the S_N V2, when the Zürich pilot observer changed from Alfred Wolfer to William Otto Brunner.

astronomical data bases: miscellaneous, Sun: activity, solar–terrestrial relations, sunspots

1 INTRODUCTION

Past solar magnetic activity has been reconstructed using the sunspot number (S_N) since 1700 and the sunspot group number (G_N) since 1610 (Hoyt & Schatten 1998; Clette et al. 2014, 2023; Clette & Lefèvre 2016; Muñoz-Jaramillo & Vaquero 2019). Over time-scales longer than the solar rotation period (about 1 month), these base indices show good correlations with other physical measures of solar activity that only exist for shorter duration, such as sunspot areas, solar radio and ultraviolet emissions, and total surface magnetic flux. Therefore, they form unique references for the solar activity and solar-terrestrial environments on a centennial time-scale (Lockwood 2013; Hathaway 2015; Tapping & Morgan 2017; Clette et al. 2021, 2023; Usoskin 2023), benefitting multiple research fields such as the physical modelling of the solar dynamo (Hotta et al. 2019; Charbonneau 2020), solar eruptions (Hayakawa et al. 2023b; Toriumi & Wang 2019), solar-terrestrial environment (Lockwood 2013; Kusano 2023), space weather (Hayakawa et al. 2019; Temmer 2021), space climate (Muscheler et al. 2016; Usoskin 2023), solar-cycle predictions (Upton & Hathaway 2018; Petrovay 2020), and solar forcing on Earth climate (Anet et al. 2014; Owens et al. 2017).

The S_N and G_N are derived from visual sunspot counts made by a large number of observers, since the beginning of the telescopic sunspot observations in 1610, either at professional observatories or by individual astronomers (Clette et al. 2014, 2023; Vaquero et al. 2016; Arlt & Vaquero 2020). The combination of those diverse observations over widely spaced epochs leads to significant discrepancies and uncertainties (Muñoz-Jaramillo & Vaquero 2019). This is particularly true for the historical parts, when observing methods were not standardized and data were sparse. Since 2011, the need for improving the homogeneity of the S_N and G_N series motivated an ongoing effort to recalibrate those key time series, starting from the original raw data (Clette & Lefèvre 2016; Cliver & Ling 2016; Svalgaard & Schatten 2016; Usoskin et al. 2016, 2021; Chatzistergos et al. 2017; Muñoz-Jaramillo & Vaquero 2019; Velasco Herrera et al. 2022; Clette et al. 2023). In 2015, this led to the release of the first recalibrated version of the international sunspot number by the World Data Center (WDC) ‘Sunspot Index and Long-Term Solar Observations’ (SILSO). This data set is the current benchmark version, hereafter abbreviated as S_N V2, or for short S_N (Clette & Lefèvre 2016; Clette et al. 2023).

In this framework, observations made over long duration by individual observers can play particularly important roles in the S_N recalibration, owing to their potentially high homogeneity, in contrast with the institutional observatories, which often involve multiple observers working in shifts, evolving teams, and sometimes also instrumental upgrades (Clette et al. 2014; Svalgaard & Schatten 2016; Mathieu et al. 2019). Remarkable examples of such dedicated observers are Sergio Cortesi (Locarno, Switzerland) and Hisako Koyama (Tokyo, Japan), who served as key references in several studies for the second half of the 20th century (Clette et al. 2014; Svalgaard & Schatten 2016). Analyses of their archival materials still continue presently (Clette et al. 2016; Cortesi et al. 2016; Hayakawa et al. 2020, 2023a). The attention to such individual observers is also essential because their source data are often difficult to access for the solar–physics community, and can easily get lost in private family archives (Clette et al. 2014; Clette & Lefèvre 2016; Carrasco et al. 2019). Moreover, we can infer their individual data stability and scaling factors in comparison with contemporaneous observers, improving the knowledge of the overall consistency of the base material leading to the S_N and G_N series (Mathieu et al. 2019; Hayakawa et al.2020, 2023a; Bhattacharya et al.2021, 2023).

This is especially important for the S_N calculation over the period 1919–1944. This interval falls in the 1849–1981 era when the Zürich Observatory was leading the production of this index (e.g. Clette et al. 2023). All the raw source counts used in Zürich were published in their journal only until 1918, not only for their internal numbers but also for numbers from multiple external observing stations. For all years between 1919 and the closing of the Zürich Observatory in 1981, most of those source data were only kept in paper archives, which had unfortunately been missing after the Zürich Observatory was closed. Fortunately, the whole part of those archives spanning the period 1945–1980 was recently recovered (Clette et al. 2021). Since then, the Eidgenössische Technische Hochschule Zürich (ETH Zürich) Library in Zürich has gathered, catalogued, and entirely scanned those handwritten tables, and the SILSO team has now undertaken the full encoding of those tables to machine-readable format in the context of the Belgian Findability and Accessibility of historical Raw Sunspot Numbers (FARSUN) project (2023–2026). For now, this thus leaves a big gap over the interval 1919–1944, during which only the raw numbers from the Zürich team itself and fragments of only a few external observers were published. While this provides the core data from which the Zürich Observatory actually produced the heritage S_N series, it prevents us from using their much broader collection of auxiliary data, in order to diagnose and improve this heritage S_N series using advanced statistical methods available nowadays.

In this context, Katsue Misawa's sunspot observations form a unique independent resource. Misawa made sunspot observations in 1921–1934, as first modern sunspot observer in Japan (Fujimori 1971; Kaneko 2001) following a few much earlier historical Japanese sunspot observers (see Yamamoto 1935; Tomita et al. 1998; Hayakawa et al. 2018a, b; Fujiyama et al. 2019). So far, Misawa's original data had not yet been identified or registered in existing sunspot data bases (Clette et al.2014, 2021, 2023; Vaquero et al. 2016), despite Kaneko's initial tabulation efforts (Kaneko 2001). Therefore, in this article, we introduce the digitization and a first survey of Misawa's original sunspot drawings and monthly numerical tables. We also present a first overall assessment of this new data set in comparison with the S_N V2 series, for the total number (Clette & Lefèvre 2016), and with the extended hemispheric SNs reconstructed from the raw hemispheric counts in the photographic catalogue of the Royal Greenwich Observatory calibrated to the S_N V2 base reference (Veronig et al. 2021).

In this article, we first provide Misawa's personal profile in Section 2. We then discuss the data and source documents in Section 3, clarify the instrumental documentations in Section 4, consider the observations and data in Section 5, and finally, we diagnose and interpret the data stability in Section 6.

2 MISAWA'S PERSONAL PROFILE

Kastue Misawa (1885–1937; Fig. 1a) was born in a farming family at Samizu village in Nagano Prefecture (N36°46′, E138°15′). After graduating from Minochi Elementary School, he continued his studies and farm works until he became an assistant teacher at Kofu Elementary School. In 1907, he obtained a teaching licence and began to teach in elementary schools in Nagano Prefectures. In 1920, he started working at Former Suwa Junior High School (presently, Nagano Prefectural Suwa Seiryo Senior High School) and taught several subjects such as geography. He also took part in research and wrote books and articles on geography, education, mineralogy, and astronomy (e.g. Misawa 1931, 1936, 1937). He taught his students while emphasizing the importance of field works. Some of his students became eminent astronomers, as exemplified by Masaaki Huruhata, who later became a director of Tokyo Astronomical Observatory.

Figure 1.

Katsue Misawa's photographs during his standard sunspot observation with his students (above) and by himself (below), reproduced with courtesy of ©︎ Misawa Family.

As recollected in Tenmon Dokokai (1922, pp. 40–43) and Misawa (1936), Misawa began his sunspot observations in early 1921 June, upon starting class at Suwa Junior High School. He had become especially interested in several large sunspots and sent his sunspot drawings to Issei Yamamoto (1889–1959) at the Kwasan Observatory of Kyoto Imperial University. Yamamoto encouraged Misawa to continue sunspot observations and even visited him at Suwa for a short discussion and supervision in early 1921 October. At that time, no individual Japanese astronomer had ever conducted such systematic sunspot observations, and Japanese astronomers thus had to rely on the Wolf number from Zürich Observatory. Misawa was significantly motivated and started regular sunspot observations from 1921 to 1934 (Fig. 1b). His monthly contributions have formed a standard sunspot data set for Japan. He only abandoned his sunspot observations at the end of 1934, due to a cataract developing in his right eye (Tenmon Dokokai 1935, pp. 168–169).

3 DATA AND SOURCE DOCUMENTS

Misawa conducted daily sunspot observations from 1921 to 1934. After his death in 1937, Misawa's sunspot records were donated to Suwa Junior High School (presently, Suwa Seiryo High School in Suwa City) along with his collections. These records are currently preserved at the (Misawa Memorial Archives 1965). These archives also host all his original sunspot drawings (Table 1), of which an example is shown in Fig. 2. We have scanned all his sunspot drawings and made them publicly available online (DOI: 10.18999/2008123).

Figure 2.

Example photographs of sunspot drawings in the Misawa Memorial Archives.

Table 1.

References for Misawa's sunspot drawings and notebooks in the Misawa Memorial Archives.

MS Shelf Mark	Titles
D-b-02	Reports of Solar Observations in 1923–1925
D-b-08	Notes on Solar Observations
D-b-11	Field Notes on Solar Observations
D-b-16	Field Notes on Solar Observations in 1922: Jan, June, and Dec
D-b-17	Field Notes on Solar Observations in 1923: Aug – Dec
D-b-18	Field Notes on Solar Observations in 1924: Jan, Apr, and May – Dec
D-b-19	Field Notes on Solar Observations in 1925: Jan and April – Sep
D-b-20	Field Notes on Solar Observations in 1925: Oct and Dec
D-b-20	Field Notes on Solar Observations in 1926: Jan – Nov
D-b-21	Records of Solar Observations in 1926: Dec
D-b-22	Records of Solar Observations in 1927: all
D-b-23	Records of Solar Observations in 1928: Oct – Dec
D-b-24	Records of Solar Observations in 1929: all
D-b-25	Records of Solar Observations in 1930: all
D-b-26	Records of Solar Observations in 1931: all
D-b-27	Records of Solar Observations in 1932: all
D-b-28	Records of Solar Observations in 1933: all
D-b-29	Records of Solar Observations in 1934: all
D-b-30	Records of Solar Observations in 1935: all

Apart from his sunspot drawings, Misawa's observational data were published as monthly tabulated summaries in Kwasan Bulletin and Tenkai (Kwasan Observatory 1922, 1923, 1924, 1925, 1926, 1927, 1928, 1930, 1931, 1932, 1933, 1934, 1935; Tenmon Dokokai 1924, 1925), as shown in Fig. 3. We consulted the monthly summary tables at the Misawa Memorial Archives while incorporating Misawa's own handwritten corrections. We manually encoded all numbers and cross-checked them with help from multiple staff members (DOI: 10.18999/2007776), as it is sometimes challenging to apply optical character recognition to numerical tables with handwritten annotations (e.g. World Meteorological Organisation 2016).

Figure 3.

Examples of Misawa's monthly summary tables in the Kwasan Bulletin of 1926 June (Kwasan Observatory 1926, Nr. 81) and in Tenkai for 1923 May (Tenmon Dokokai 1923, p. 426).

We generated the data files for each year, and three files specifically for 1925. We then integrated them into three files according to the different data formats (DOI: 10.18999/2007776). This proved necessary because Misawa changed his summary format several times. From 1921 October to 1925 June, Misawa recorded the numbers of sunspot groups (Group), large individual sunspots (LS), other individual sunspots (Spot), and faculae with his notes (e.g. Fig. 3b), defining ‘LS’ as sunspots with penumbrae (Tenmon Dokokai 1922, p. 43; Fig. 4). In 1925 July–August, Misawa stopped recording LS separately and instead start separating facula numbers into Eastern and Western solar hemispheres. From 1925 September to 1934 December, Misawa recorded individual SNs (including LS) for each sunspot group separately in the northern and southern solar hemispheres (e.g. Fig. 3a).

Figure 4.

Examples of large sunspots (LS) and other sunspots on 1930 October 10, (09:30 LT) and 1929 November 5 (08:40 LT), as reproduced from Misawa (1936, p. 119). LS numbers are shown in brackets, in contrast with numbers for normal sunspots without brackets.

4 INSTRUMENTAL DOCUMENTATIONS

Katsue Misawa conducted his sunspot observations mainly at Suwa Junior High School (N36°02′, E138°08′). Misawa used a 3-in refractor with a magnification of × 83 throughout his career. He made eyepiece observations with direct vision through filtering glass (Misawa 1936, p. 139).

He at least once borrowed telescopes with projection boards from Issei Yamamoto of Kyoto Imperial University (Fig. 1). He also knew how to observe sunspots by projecting the solar image on a paper screen. This was because he found it difficult to use a projection screen (Misawa 1936, pp. 132–144). Indeed, his light-weight set-up was probably not rigid enough for stable pointing, and the quick image drift experienced with an undriven mount was incompatible with the typical duration of several minutes needed to mark all sunspots at their exact position. This 3-in telescope has been preserved at the Misawa Memorial Archives, as shown in Fig. 5. The 3-in telescope aperture has been described differently, i.e. as 3 in in 1921–1923, 8 cm in 1925–1929, 7 cm in 1929–1930, and 73 mm in 1930–1934 in Kwasan Bulletin, owing to unit conversion issues.

Figure 5.

Misawa's 3-in refractor telescope, preserved at Misawa Memorial Archives. Misawa's original sunspot drawings are shown behind this telescope, generally bundled for each month.

Seemingly, Misawa's drawings were only approximate sketches showing the rough location of each group on the solar disc, based on visual proportions seen at the eyepiece in the aerial image. The aspect of each group, shown in separate close-up views in the margins, must be considered as an accurate rendition of the morphology of each group, with a complete census of all visible spots, but with only approximate proportions. Therefore, those sketches are fully usable for accurate counts of sunspot groups and individual sunspots and for the morphological group classification, whereas sunspot areas cannot be extracted reliably unlike real drawings made by projection.

Like in most telescopes, Misawa's telescope and eyepiece combination showed inverted images. Here, we compare two such examples: Misawa's sunspot drawings on 1929 December 12 and 21, (Fig. 6) and those on the same dates from the Mt. Wilson Observatory with side-corrections (Fig. 7). Misawa reported four and five groups in the Northern and Southern solar hemispheres on 1929 December 12 (Kwasan Observatory 1929), respectively; Misawa's sunspot drawings on this date (Fig. 6a) show five and four groups in the upper and lower halves of the solar disc, respectively. Likewise, Misawa reported nine and four groups in the Northern and Southern solar hemispheres on 1929 December 21 (Kwasan Observatory 1929), respectively; Misawa's sunspot drawings on this date (Fig. 6b) show four and nine groups in the upper and lower halves of the solar disc, respectively. These examples show that Misawa considered the upper and lower halves of the solar disc as the Southern and Northern solar hemispheres, respectively. Moreover, their general configurations match well with those shown in the orientation-corrected sunspot drawings from Mt. Wilson on the same dates (Fig. 7) when Misawa's drawing is rotated 180° from the original form. These examples confirm that Misawa's sunspot drawings were shown in inverted images through his telescope eyepiece, with reasonable identifications for the solar equator.

Figure 6.

Misawa's published sunspot drawings on 1929 December 12 (12:00 LT) and 1929 December 21 (11:30 LT) with his annotations and group classifications, reproduced from Misawa (1936, pp. 138 and 142). The dashed lines separate the neighbourhood of the solar equator from the high-latitude zones in each solar hemisphere. They mark the orientation of the solar equator, as determined by Misawa, and vary according to the tilt of the solar axis versus celestial North (P angle), and to the local time and date of the observation, given the field rotation due to the azimuthal mount (tilt of celestial North).

Figure 7.

Sunspot drawings on 1929 December 12 (above) and 1929 December 21 (below) from © Mt Wilson Observatory. These images have been reversed horizontally to match the true orientation of the solar disc as seen in the sky, as taken from Baranyi, Győri & Ludmány (2016) and Győri, Ludmány & Baranyi (2017).

Misawa conducted most of his observations by himself, although his students occasionally assisted his observations. Consulting Misawa's monthly summaries, we identified at least four assistants throughout his observations: Masaaki Huruhata (古畑正秋: 1912–1988), Kazuaki Gomi (五味一明: 1910–1990), Yoshihiko Kosai (河西慶彦: 1903–1961), and Katsumi Yamaoka (山岡克巳: 1903?–1958). These students later contributed to Japanese astronomy, as shown in Fujimori (1971). The monthly reports were annotated with one of Misawa's assistant's names when one of them made sunspot observations instead of Misawa himself.

5 OBSERVATIONS AND DATA

We scanned Misawa's sunspot drawings at the Misawa Memorial Archives (DOI: 10.18999/2008123) and Misawa's monthly summary tables for his sunspot counts in the Kwasan Bulletin and Tenkai. These records allowed us to derive 4032 d of sunspot observations from 1921 October to 1934 December, thus corresponding to a particularly high productivity of 25.4 days per month on average (i.e. 83 per cent; Fig. 8). Among them, only 366 d of observations were associated with his assistant observers (blue bars in Fig. 8), in contrast with Misawa's own observations for the other days (red bars in Fig. 8). In fact, Misawa observed the Sun slightly more frequently than the Zürich Observatory (Misawa 1936, p. 109), which seems to be due to a better temporal coverage during the winter season in comparison with that of Zürich (cf. Section 6.2 and Fig. 13). Given the vagaries of the daily weather, this indicates his full dedication to this observing task.

Figure 8.

Frequency of Misawa's sunspot observations (days per month). The red bars and blue bars indicate observational days associated with Misawa himself and with his assistant observers, respectively.

For a first comparison, we plotted together the monthly means of Misawa's raw daily Wolf numbers (W_M = 10 G_M + F_M)¹ over his whole observing career, 1921 October–1934 December, and S_N (Fig. 9), which closely corresponds to the Zürich raw Wolf numbers. Indeed, over this entire period, the Zürich Observatory used as primary S_N value its own raw Wolf numbers for all days when the Sun could be properly observed in Zürich (about 80 per cent of all days; Clette et al.2014, Clette & Lefèvre 2016), and S_N V2 still equals this original Zürich series for all times before 1981. Overall, we can see that W_M is systematically lower than S_N, but with very similar variations. In Fig. 10, we plot the monthly mean W_M versus S_N V2, as well as a linear regression. The latter shows very good linear correlation (correlation coefficient R = 0.972), with a correction factor of 0.726.

Figure 9.

Misawa's raw Wolf numbers W_M (red curve) and S_N V2 (green dashed curve) produced by WDC–SILSO (Clette & Lefèvre 2016).

Figure 10.

Scatterplot for comparing Misawa's monthly mean Wolf number and the monthly mean value of S_N.

Misawa also recorded the hemispheric SNs from 1925 October to 1934 December. We computed Misawa's hemispheric asymmetry index [HA = (W_NH − W_SH)/(W_NH + W_SH)] based on Misawa's monthly reports, and the same asymmetry index for the extended hemispheric numbers from Veronig et al. (2021), as shown in Fig. 11. While the match is almost perfect after 1931, the asymmetry given by Misawa is much smaller in the earlier part, in particular before 1928. This could indicate a problem in the proper determination of the position and orientation of the solar equator in Misawa's drawings (P and B angles, azimuthal field rotation; cf. Section 4, Fig. 6), leading to a random mixing of groups from the two hemispheres, and thus a drop of the hemispheric asymmetry. The very good match in the final years could indicate that Misawa may have corrected, in one or two steps (1928 and 1930), shortcomings in his initial drawing-orientation method.

Figure 11.

Hemispheric asymmetry indices [HA = (R_NH − R_SH)/(R_NH + R_SH)] of Misawa's sunspot observations (Misawa HA) in a red curve and hemispheric SNs (V + 21 HA; Veronig et al. 2021) in a green broken curve.

6 TEMPORAL DATA HOMOGENEITY

6.1 Comparison with S_N

Fig. 12 (upper panel) shows the temporal evolution of the monthly mean ratio between Misawa's Wolf numbers and S_N. Misawa's data appear relatively stable over the entire series, with an overall mean scale factor of 0.73 ± 0.16 relative to S_N. In particular, no significant trend is found between the first and last observations of Misawa's time series, indicating the absence of a significant impact from the eyesight degradation due to ageing. Indeed, the mean scaling factors are respectively 0.79 ± 0.23 over 1921–1924 and 0.76 ± 0.12 over 1929–1934, which thus matches closely, well within the uncertainties.

Figure 12.

Temporal evolution of Misawa's Wolf numbers W_M relative to SILSO S_N V2. The upper panel shows the temporal evolution of the monthly mean W_M/S_N ratio. The error margins (light-blue shading) correspond to W_M/(S_N ± σ_SN), where σ_SN is the standard error on the monthly mean S_N. The lower panel shows the monthly mean differences W_M–S_N after dividing W_M by the mean ratio 0.73 ± 0.16, for the entire duration of the series 1921–1934, in order to normalize the scale. The error margins are also corresponding to the standard error on the monthly mean S_N.

We then rescaled Misawa's data by the above mean scaling factor and computed the difference with S_N (lower panel of Fig. 12). Those differences reveal a strong downward drift over the 3-yr interval 1925–1928, which also appears in the ratios. We note in particular the sharp drop in 1925 September, which corresponds to the change in recorded data in Misawa's monthly tables (Section 3), suggesting that this methodological transition may have had a direct influence on Misawa's spot counts.

We also observe two other significant drops in the ratio at the very beginning (1921 October to 1922 March) and very end of Misawa's observations (1934 February to 1934 October). Those brief intervals correspond to the final decline of cycle 15 (cycle minimum in 1934 August; cf. https://www.sidc.be/silso/cyclesminmax) and the early rise of cycle 17 (minimum 1933 September). The very low values around the two minima (S_N<20), lead to larger fluctuations of the ratio during the intervals 1922–1924 and 1931–1934, but also to very large errors, thus making those variations insignificant. This is confirmed by the very low differences over those same time intervals (lower panel in Fig. 12). Between 1924 and 1930, the ratio also drops as solar cycle 16 reaches its peak (maximum in 1928 April).

Overall, although Misawa only caught short sections of cycles 15 and 17, those fluctuations thus suggest a solar cycle dependency of the W_M/S_N ratio. However, the actual drift does not exactly mirror the cycle 16 modulation. Indeed, this dependency does not hold in the early rise of cycle 16, from 1923 to mid-1925, before the sharp methodological change occurs, and the ratio and differences return to their mean value already at the end of 1928 (with some strong fluctuations during that year), thus soon after the maximum of cycle 16, and for the rest of the declining phase. Anyway, we may wonder what caused this underestimate, which grew as solar activity increased.

A first explanation may reside in Misawa's non-standard counting rule, separating large and small spots (Section 3). Indeed, contrary to Wolfer's standard rule used in Zürich since 1893, where all umbral cores must be counted as separate sunspots (Clette et al. 2014), Misawa counted a penumbra containing multiple umbrae as one LS, while all isolated small sunspots without penumbra were counted individually. For instance, on 1930 October 10, Misawa counted 42 small spots and 8 penumbrae (LSs) in the group of Fig. 4 (left panel), leading to a total count of 50 spots, whereas we get a total of 57 when counting all individual umbrae (thus, an undercounting ratio 0.88 for spot counts). On 1929 November 5, Misawa counted 35 small spots and 3 LSs in the group of Fig. 4 (right panel), giving a total of 38, whereas we count 48 individual umbrae (ratio of 0.79 for spot counts). Similar trends are confirmed in Misawa's whole-disc sunspot drawings (Fig. 6). Here, Misawa counted 9 groups and 66 spots on 1929 December 12 and 13 groups and 49 spots on 1929 December 21 (Fig. 6), whereas we would count them as 9 groups and 86 spots on 1929 December 12 (ratio of 0.88 for the Wolf number) and 13 groups and 60 spots on 1929 December 12 (ratio of 0.94 for the Wolf number).

These examples show that Misawa's counting method inevitably underestimated the individual sunspot count and thus also the Wolf number, relative to the standard Zürich S_N. Moreover, as the fraction of large spots is low near solar cycle minima and increases at high activity, this undercounting will amplify with solar activity, which matches the variations observed in Misawa's scale factor. However, we note that the above 0.8–0.9 deficit falls short of the mean 0.73 scale factor for Misawa's W_M, indicating that another factor is probably also playing a role. We also note that Misawa's accurate drawings allow a full recounting of individual sunspots according to the standard Zürich rules, which should allow eliminating this discrepancy and aligning Misawa's numbers with the international standard.

6.2 Comparison with three individual Zürich observers

Pushing further our diagnostics, we considered the separate raw data series from individual Zürich observers who were active during Misawa's observing career, as their data were published and are thus available, and as they include separate group counts and sunspot counts, next to the Wolf numbers. Fig. 13 gives the chronology of the monthly number of observations for Misawa and three Zürich observers: pilot observers Alfred Wolfer and William Otto Brunner and an assistant observer Max Broger. It turns out that this time interval includes the transition between two primary pilot observers, Wolfer and Brunner, in 1926–1928, i.e. the period between the arrival of Brunner and the end of Wolfer's observations (Dudok de Wit et al. 2016; Clette et al. 2021).

Figure 13.

Chronology of the number of observing days per month for Misawa et al. from Fig. 8, and three Zürich observers: two pilot observers (Wolfer and Brunner), and one long-term assistant observer (Max Broger), as derived from Clette et al. (2021). We observe that while the Zürich temporal coverage shows a marked drop during winters, Misawa's coverage hardly shows significant seasonal variation.

This figure first shows that Misawa's monthly number of observing days was generally larger than Wolfer's or Brunner's, especially during the boreal winters, probably owing to the humid weather conditions during Swiss winters. Moreover, once Brunner started his observations in early 1926, Wolfer's monthly coverage decreased globally, most probably because Wolfer shared part of the observing duties with Brunner and he continued observing in parallel with him mainly to allow direct comparisons and a seamless transition between himself and his successor. Alfred Wolfer himself mentioned that his observations were exceptionally incomplete in 1925, owing to his absences on July 10–29 and September 8–26 and his sick leave from September 29 to November 23 (Wolfer 1927, pp. 153–154 and 163). As the S_N for that period was largely based on the raw Wolf numbers from the Zürich pilot observers, the 1926–1928 (Wolfer-Brunner) transition may have left its marks in the resulting S_N series. Conversely, as Misawa's data are bridging the Wolfer and Brunner periods, they may also provide precious independent information about the stability of the S_N series across this transition.

In order to resolve this problem, we have compared Misawa's data with all three Zürich observers, separately in terms of Wolf numbers (W; Fig. 14), sunspot group count (N_g; Fig. 15), and individual sunspot count (N_s; Fig. 16). These figures show that the downward drift found in the comparison with S_N over 1925–1928 (Fig. 13) appears in all three ratios.

Figure 14.

Temporal evolution of the monthly mean ratio of Wolf numbers (W_M) from Misawa versus those from Wolfer, Broger, and Brunner. The shaded error range indicates the standard deviations of the daily ratios in each month.

Figure 15.

Temporal evolution of monthly mean ratios of group counts (N_g) from Misawa versus those from Wolfer, Broger, and Brunner. The shaded error range indicates the standard deviations of daily ratios in each month.

Figure 16.

Temporal evolution of monthly mean ratios of sunspot counts (N_s) from Misawa versus those from Wolfer, Broger, and Brunner. The error margins indicate their standard deviations of the daily ratios in each month.

Fig. 14 for the Wolf number closely corresponds to the comparison with S_N, which is closely tracking the pilot observers. The uncertainties are larger for those individual observers, mainly due to the incomplete monthly coverage. We also observe that the ratios for the three Zürich observers match extremely well, within those uncertainties, in spite of the fact that their observing days do not always match within each month. This conclusion is also valid separately for N_g and N_s. Stronger deviations occur for a few isolated months, and correspond to months when an observer had very few observing days, as indicated by the larger standard deviation.

Looking now at the separate counts, the drop affecting N_s (Fig. 16) matches the undercounts due to Misawa's use of the non-standard LS category, as illustrated above through a couple of sample data and drawings. Interestingly, during the cycle 15–16 minimum, over 1923–1924, Broger tends to count more spots than Wolfer, although the difference was actually small in comparison with the large relative uncertainties, given the low S_N values during that period. As there is no corresponding difference in the group counts N_g, this N_s deficit affects to a lower extent Wolfer's Wolf numbers. This excess in Broger's numbers when activity is low was already identified as a consequence of the use of weighted counts by this Zürich assistant (see fig. 19 in Svalgaard, Cagnotti & Cortesi 2017).

Overall, those comparisons with individual Zürich observers show almost exactly the same results for all of them, and for the S_N. However, this should not lead to the conclusion that the 1925–1928 deviation must be attributed to a flaw in Misawa's numbers, simply because he is the common term of all three comparisons. Indeed, we know that in the Zürich system, all Zürich observers were assumed to count sunspots in the same way as the pilot observer, and all were attributed the same personal k coefficient of 1. Therefore, assistants were trained to keep track of their personal counts on a monthly basis relative to the pilot observer, and to adapt their counting practice to adjust accordingly. In fact, a modified counting method, attributing different weights to each spot according to its size, was implemented by Wolfer in the early 20th century for helping assistants to achieve this self-normalization (Waldmeier 1948; Cortesi et al. 2016; Svalgaard, Cagnotti & Cortesi 2017). Therefore, we can expect matching results for those three individual comparisons, which are thus not at all independent. Consequently, this comparison confirms the expected consistency among the Zürich observers, but does not allow us to determine reliably if the problem resides in Misawa's series or in the 1925–1928 transition in Zürich, or perhaps even in both series.

7 SUMMARY AND DISCUSSION

In this first presentation of the newly digitized data from Katsue Misawa, we could establish a rather clear, though not fully complete, picture of his instrumentation and observing practices. This very dedicated observer (>25 observing days per month) was producing his drawings based on direct observations through the eyepiece, and not by projection on a screen, as indicated by the reverse and not mirrored orientation of those drawings. Those drawings thus consist in whole-disc sketches showing the relative position of all sunspot groups, with detailed close-up views of each group drawn in the margins.

Misawa's archives also host monthly summary tables providing Misawa's original sunspot and group counts. We found that Masiwa changed several times the count formats in particular in 1925, and his resulting daily Wolf numbers W_M are on average 27 per cent lower than the international sunspot number S_N. Particularly in most of his counts, Misawa made a distinction between LSs with penumbrae and small spots without penumbra. By comparing Misawa's drawings with simultaneous drawings from the Mt Wilson Observatory, and by computing the ratio between his sunspot counts with those from observers of the Zürich Observatory, we can conclude that this deficit is partly attributable to the fact that Misawa counted all LSs as one, regardless of the number of separate umbral cores within the same penumbrae, contrary to the standard counting rule in application for the S_N reference series. By a comparative analysis of the hemispheric numbers produced by Misawa only after 1925, we find that the variation of his HA index is significantly lower than the Greenwich-based HA before 1931. This requires further investigation but may indicate an imperfect separation of hemispheres in Misawa's drawings, affected by the changing orientation of his drawings due to the use of an azimuthal mount for his telescope.

More importantly, the comparison with S_N as well as with the separate Zürich observers, indicate significant temporal variations of the W_M/S_N ratios and W_M –S_N differences, with drops (Misawa's W_M becoming lower relative to S_N) of about 96.5 per cent in the first and last years of the series, and in particular over the interval 1925 September–1928 December. Those variations are temporally aligned on the evolution of the solar cycle, from the end of cycle 15 to the onset of cycle 17. They may thus indicate a global dependency on the level of solar activity (non-linear relation between W_M and S_N), an effect expected for the above-mentioned non-standard counting of LSs in Misawa's method. However, some features, like the sharp jump in mid-1925, suggest that those variations may indicate a true inhomogeneity in the data.

In this respect, we note the coincidence of this 1925 jump with the corresponding change in the contents of Misawa's monthly tables, possibly indicating a bias due to a change in his observing strategy. However, in our comparison with individual observers in Zürich, we observe that this 1925–1928 interval also coincides with Wolfer's leaves and the transition between Wolfer and Brunner, as primary pilot observers for the Zürich numbers, which form the base of the S_N for that epoch. Overall our first comparison, which includes only Misawa and Zürich, does not allow to determine which of the series may be affected by these deviations. Still, as the deviations are found both in the separate group and sunspot counts, we can conclude that the problem cannot reside only in Misawa's non-standard counts for LSs. The subdivision in groups must also play a role. For now, we can at least observe that this was a temporary deviation, without global monotonous scale drift, as the scales match at the beginning and at the end of the 1921–1934 time window.

Therefore, this leaves us with three possible causes for the 1925–1928 discrepancy: a drift in Misawa's numbers due to an undocumented change in his observing practices or a deviation in the S_N due to an unsuspected temporary side effect of the Wolfer–Brunner transition occurring in the same time frame, or their combination. In order to find a more robust answer, the next step would be to exploit more in detail Misawa's original documents. In particular, his drawings can give us a direct clue on how he was splitting the groups. In addition, the detailed and accurate close-up sketches of individual groups would allow a full recounting of individual sunspots and umbrae, thus replacing the non-standard counts reported by Misawa in his monthly tables.

Regarding the Zürich numbers, a very important but thus-far unexploited resource are the personal logbooks of Wolfer, Brunner and their assistants, which include sketches and tables showing the groups and the number of spots counted in each individual group using the standard 80 mm refractor. Like Misawa's original drawings, they fully document how the individual groups and sunspots were counted to produce the daily total numbers reported in the monthly and yearly tables. This collection is preserved in the archives of the Library of the ETH (Catalogue entry: Hs 1050:123 to 218), and has recently been entirely scanned (accessible online at e-manuscripta.ch). Those 96 logbooks are new essential keys to the observing practice of each individual observer (A. Wolfer, W.O. Brunner, M. Broger, F. Buser, W. Brunner, and B. Beck) from 1881 to 1947, and can thus help diagnosing a possible transitional bias between Wolfer and Brunner over 1925–1928 and thus clarify the discrepancy with Misawa's numbers.

Given the size of those collections, those further analyses go beyond the scope of this first article. For now, the new Misawa data series proves to be doubly valuable. First, it currently helps to fill the 1919–1944 gap in the Zürich sunspot archives, and if those documents are hopefully rediscovered one day, it will anyway enrich this base collection with high-quality data that were not collected by the Zürich Observatory. Moreover, it so happens that Misawa's observing period fully overlaps one of the four main transitions that took place between Zürich pilot observers in the history of the S_N (Clette et al. 2021). As those pilot observers provided the base reference for the production of S_N, Misawa's brings an independent diagnostic and validation tool for the homogeneity of the S_N, similarly to the much earlier Schwabe–Wolf transition studied in Bhattacharya et al. (2023). Finally, the recovery of Misawa's data series shows that long-forgotten series can still be rediscovered, and should motivate further efforts to recover more observers to fill the 1919–1944 data gap, in order to ultimately achieve a multi-observer S_N reconstruction.

ACKNOWLEDGEMENTS

We thank the Misawa Family, the Memorial Archives, and Nagano Prefectural Suwa Seiryo Senior High School for letting us access, research, and publicize Misawa's sunspot drawings. We thank Mt. Wilson Observatory and WDC-SILSO for providing the magnetograms and the international sunspot number. This research was supported by NIHU Multidisciplinary Collaborative Research Projects NINJAL unit ‘Rediscovery of Citizen Science Culture in the Regions and Today’ (H421042227). HH thanks Mayuko Watanabe, Satoshi Nozawa, and Takuichiro Ohnishi for their intensive help to access Misawa's archival collections and background information, and Momoko Hattori for her help to encode Misawa's tabulated data sets. This research was conducted under the financial supports of JSPS grant-in-aids JP20K20918, JP20H05643,JP21K13957, and JP22K02956. HH has been part funded by JSPS Overseas Challenge Program for Young Researchers, the ISEE director's leadership fund for FYs 2021–2022, Young Leader Cultivation (YLC) programme of Nagoya University, Tokai Pathways to Global Excellence (Nagoya University) of the Strategic Professional Development Program for Young Researchers (MEXT), and the young researcher units for the advancement of new and undeveloped fields, Institute for Advanced Research, Nagoya University of the Program for Promoting the Enhancement of Research Universities. HH acknowledges the International Space Science Institute and the supported International Teams #510 (SEESUP Solar Extreme Events: Setting Up a Paradigm), #475 (Modeling Space Weather And Total Solar Irradiance Over The Past Century), and #417 (Recalibration of the Sunspot Number Series). FC, SB and LL acknowledge support by the Belgian Solar Terrestrial Center of Excellence (STCE, BelSpo).

DATA AVAILABILITY

The Misawa sunspot records are preserved in the Misawa Memorial Archives. Misawa's sunspot drawings are available at DOI: 10.18999/2008123. Our data for Misawa's sunspot number is available at DOI: 10.18999/2007776. We acquired Mt. Wilson magnetograms from Mt. Wilson Observatory. Their exact references are given in Table 1. The international sunspot number is provided from the WDC–SILSO.

Footnotes

1

Here, we use G_M and F_M to describe Misawa's group number and individual spot number.