The Extreme Space Weather Event of 1872 February: Sunspots, Magnetic Disturbance, and Auroral Displays

The Chapman–Silverman storm of 1872 February 4 occurred in the declining phase of Solar Cycle 11, which had its maximum in 1870 August (Hathaway 2015; Clette & Lefèvre 2016). Empirically, intense space weather events frequently occur during the declining phase of large solar cycles (Lefèvre et al. 2016; Owens et al. 2022). In the week preceding the event, the international sunspot number (Version 2.0; Clette et al. 2014, 2015; Clette & Lefèvre 2016) increased significantly from a value of 95 on January 28 to 272 on February 2, following the appearance on the solar disk of the active region that we identify as the source of the extreme storm, as shown in Figure 1. Archival records available at the libraries of the INAF Osservatorio Astronomico di Roma and Osservatorio Astrofisico di Catania document detailed solar observations by Angelo Secchi (1873, pp. 8–12 and 16–19) in Rome (see Secchi 1872b, p. 83; MS OAR 1–4​​​​)​​, ¹⁹ Pietro Tacchini (1872, p. 83) in Palermo, and Francesco Denza in Moncalieri (Secchi 1872b, p. 83) during 1872 (Figure 1; see footnote 19). Secchi's sunspot drawing on January 27 shows a significant increase of prominence activity on the solar limb immediately before the entrance of the group numbered 29 in Secchi's drawings (hereafter Group #29; ²⁰ see Figure 2) with his annotation "(solar) activity has re-started (È ricominciata l'attività)" (Osservatorio Astronomico di Roma, MS INAF 1, f. 21) and associates this enhanced prominence activity with Group #29 (Secchi 1872c, p. 16).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Because the structure and complexity of solar active regions are essential keys to understanding the generation of solar eruptions (Toriumi & Wang 2019), we have analyzed the series of full-disk solar drawings by Angelo Secchi ²¹ (Figure 2), Pietro Tacchini ²² (Figure 3), and Gustave L. Bernaerts ²³ (Figure 4) from their observations of the Sun in the weeks around the event. These three figures show the westward migration of the solar features to the left in the figures because the observations were performed with the telescopes in projection mode (mirror inversion in the east–west direction). The series of drawings of the three observers show the first appearance of sunspot Group #29 on January 28, in association with an increase in the sunspot number from 95 to 150 at that time (shown in Figure 1). These drawings also reveal a second sharp increase of the sunspot number on February 1–2 (from 188 to 272), which was mainly associated with an intrinsic growth of Group #29, whereas two small groups also came into sight at the east limb of the solar disk (see also Denza 1872, p. 825). Group #29 in Secchi's drawings reached its maximum extension on February 2 (627 μsh). On February 3, it started to shrink, with an area of 461 μsh; it was last seen near the west limb on February 9 by Bernaerts. Group #29 was thus not exceptionally large, similar to the medium-sized active region (≈600 μsh) associated with a great magnetic storm (minimum Dst* ≈ −595 nT; Love et al. 2019a) on 1909 September 25 (Hayakawa et al. 2019a). The next largest groups on February 3 (Groups #28 and #34 in Figure 2) had areas of 166 μsh and 111 μsh, respectively.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The "large group" in Meldrum (1872a) was almost certainly Secchi's Spot Group #29 in the southern hemisphere, and the "chain of sunspots" that Meldrum (1872a) mentioned corresponds to Spot Groups numbered 28, 33, 34, 27, 26, and 32 by Secchi in the northern hemisphere. Because the major magnetic storm occurred on February 4, the responsible flare and eruption likely occurred 1–2 days before the storm (i.e., February 2–3). Although Jones (1955) did not specify the source sunspot group for this storm, Group #29 seems to be the most plausible origin of the source flare because of its favorable location near the central meridian (S19° E05°) and its relatively large size (461 μsh) on February 3. Emprically, it has long been known that intense magnetic storms originate preferentially from regions in the vicinity of the central meridian (Newton 1943; Akasofu & Yoshida 1967; Cliver et al. 1990; Gopalswamy et al. 2012; Lefèvre et al. 2016; Cliver et al. 2022b).

Group #29 had a complex and unusual shape. No clear pair of dominant spots (global magnetic dipole) can be discerned. Contrary to most sunspot groups, which are elongated in the East-West direction (dipole parallel to the solar equator), this group did not have a particular axis of symmetry. The general shape and proximity of multiple individual sunspots suggest magnetic polarity mixing and a convoluted magnetic neutral line topology inside the group, and thus greater non-potential or free energy in this active region (Zirin & Liggett 1982; Török & Kliem 2003; Toriumi et al. 2017; Bourdin & Brandenburg 2018; Bourdin et al. 2018). At least three of the spots had substantial penumbrae on February 3, which indicates that they contained a strong magnetic flux.

On February 3, Secchi added an annotation, "in a spiral" (a spirale) to this group. This spiral structure may be related to the series of new pores extending from this spot (see the bottom of Figure 2 for its evolution from February 1–5), as seen in the satellite sunspot type of Toriumi et al. (2017) and Toriumi & Takasao (2017). If this is the case, then we can find a comparable pertinent example: NOAA active region (AR) 5395 of 1989 March (Figure 1 of Wang et al. 1991), where new sunspots (e.g., P4, P5, and P6) surrounded the large existing spot (F1) and produced a clockwise outward spiraling field or peculiar vortex-like motion of satellite sunspots (Tang & Wang 1993; Ishii et al. 1998). This peculiar configuration led to very powerful X-class flares (X15.0 on March 6 and X4.5 on March 10) and to a great magnetic storm with minimum Dst = −589 nT on March 14 (Allen et al. 1989; Tang & Wang 1993; WDC Kyoto et al. 2015).

In summary, Group #29 in Secchi's drawings was most likely the source of the C-S storm, based on its rapid evolution, favorable location, relative size, and complexity. ²⁴

There is no direct evidence of a flare and solar eruption in Group #29 in Secchi's drawings. This is not surprising given that there were no systematic flare patrols in the early 1870s (Švestka & Cliver 1992) and technological limitations prevented observations of the chromosphere over the solar disk in monochromatic light. While flares were occasionally observed spectroscopically as brilliant reversals in chromospheric lines, the spectroheliograph (Hale 1892) and spectrohelioscope (Hale 1926), which permitted chromospheric imaging, had yet to be invented. A systematic flare patrol in the Hα line only started in 1934 (D'Azambuja 1934).

Nevertheless, the analyzed archival documents hint at an eruptive Sun on the days around the C-S storm given that contemporary Italian astronomers commented on enhanced solar activity observed at the solar limb with their spectroscopes on 1872 February 3–4. On February 3 Denza recorded (Denza 1872, p. 825) a significant agitation of the chromosphere and "a beautiful eruption (une belle éruption)" within "several small prominences (plusieurs petites protubérances)" at 10–11 local time (LT), viz., 9–10 UT. One of these eruptive prominences was situated "at 112°–114° from the North (à 112–114 degrés du nord)" in the southwest limb and reached "a height of 3' (la hauteur était de près de trois minutes d'arc)" (Denza 1872, p. 825). On the same day, Secchi (1872a, p. 586) recorded a "continuation of the bright mass" at S40° E10° of the solar disk around 10 LT and described a series of prominences on the west limb between 15° and 40° south, which is consistent with Denza's observation. Secchi described the prominence activity on February 3–4, as "alive and beautiful (viva e bella)" in the margin on the right-hand side of Figure 2, like "wonderful jets (belliss. fili)," and "even if one tried, they could not make [the jets] so beautiful and even (a farlo apposta non si farebbero i peli così belli e regolari)" in his solar drawings (Osservatorio Astronomico di Roma, MS INAF 1, ff. 28–29). He also reported "a mass of high and inclined jets with holes underneath (fili alti della cromosfera and massa di fili belli ad archi intrecciati con fori sotto)" in his logbook (Secchi 1872c, p. 17). Here, the "holes" might refer to a set of prominence threads, or maybe even post-flare loops. As also cited in Secchi (1872a, p. 587), Tacchini's contemporary observation highlighted "to the west, a beautiful eruption (à l'ouest, belle éruption)" on February 3.

These reports agree reasonably well with one another on the location and timing of the eruptions observed on the south-western limb on February 3. These eruptions may be indirectly related to the C-S storm, which began with a sudden commencement (SC) at ≈14:27 UT as shown in Figures 5 and 6 (Moos 1910, p. 452; Mayaud 1973; Silverman 2008) on February 4, ≈29 hr after the reported eruptive activity, which implies a mean propagation speed of the associated ICME of ≈1450 km s⁻¹. A putative identification of a source flare near disk center for the C-S storm with the prominence activity from ≈9–10 UT on February 3 is supported by both the size of the storm and the plausibly short ICME transit time of ≈29 hr for a central meridian eruptive flare.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Although the prominence observations only revealed activity at the limb, when taken together with the location of Group #29 and the timing of the onset of the great storm, this suggests the reasonable possibility that the limb activity was triggered by a major eruptive flare with an associated large-scale wave (e.g., Balasubramaniam et al. 2010). Cliver et al. (2005) reported a widespread wave (manifested by EUV dimming at the limb in an Extreme-ultraviolet Imaging Telescope (Delaboudinière et al. 1995) difference image) from a near-disk-center backside flare that was associated with a high-energy proton event in 2001 August (Cliver et al. 2005). From Figure B1 in Cliver et al. (2022b), which relates the maximum CME speed at Earth to the Sun-Earth transit time of the ICME (≈ 29 hr in this case), the likely ICME arrival speed at Earth was ≈1100 km s⁻¹.

The storm SC indicated that the ICME-driven shock (followed by the shock-sheath structure and fast CME body) passed Earth beginning at ≈ 14:27 UT on 1872 February 4. The storm, recorded at the Royal Observatory Greenwich (GRW: N51°29', E000°00', 546 MLAT; Royal Observatory Greenwich 1872; see also Valach et al. 2019), ²⁵ was violent but of relatively short duration (≈22 hr; Jones 1955, p. 105). Figure 5 shows the digitized H and Z traces of the magnetogram from Greenwich Observatory, which was in the afternoon sector during storm onset.

Although the traces are greatly disturbed, some aspects of the storm phases can be discerned (see also Valach et al. 2019). After the shock arrival at 14:27 UT, ΔH shows a positive excursion lasting to just after 16 UT. This is consistent with a sudden commencement during an initial storm phase. In a statistical study, Shinbori et al. (2009) showed that the ΔH amplitude at St. Paratunka (46° MLAT) is 2.3 times larger than the SYM-H index at 15–16 MLT. The maximum amplitude of this positive excursion is ≈200–250 nT, on average, and the peak ΔH is ≈72 nT. The ratio is ≈ 2.7–3.5, which is comparable to the result of Shinbori et al. (2009). The ΔH amplification at mid-latitude can be explained by the contribution from the ionospheric current (Araki 1994; Shinbori et al. 2009). Subsequently there was an interval with large jerks in the Greenwich magnetogram components that lasted until ≈18:30 UT, which was followed by a negative excursion from ≈18:35 to 19:05 UT and a positive recovery (with overshoot) from ≈19:05 to ≈20:30 UT.

The oscillations between 15:00 and 16:00 UT and after 22:30 UT are notable features of the magnetogram traces in Figure 5. These low-frequency oscillations seem to be Pc-5 (Pulsation continuous-5) signals, which appear occasionally in the storm main phase and often in the recovery phase of intense geomagnetic storms (e.g., Huang 2021). Such pulsations often register in the auroral region but can also occur simultaneously at all latitudes and longitudes/local times, with a period between a few to 10 minutes. Periodic solar wind density structures are an important, but not exclusive, source of the fluctuations (Di Matteo et al. 2022). Low-latitude giant pulsations were reported in the recovery phase of the 1972 August 4–5 storm (See Figure 1(e) of Knipp et al. 2018) when solar wind speed exceeded 1100 km s⁻¹ (Vaisberg & Zastenker 1976). Heyns et al. (2021) used mid-latitude measurements to show that Pc-5 pulsations can drive significant Geomagnetically Induced Currents (GICs) at a mid-latitude network during intense geomagnetic storms for an extended duration. During the storm recovery phase on 2003 October 31, Pc-5 pulsations were noted in the mid-latitude Czech pipelines. During much of that day the solar wind flow speed exceeded 1000 km s⁻¹. Other researchers (Sakurai & Tonegawa 2005) reported the 2003 October Pc-5 pulsations as some of the largest ever recorded in the Pc-5 band. Owing to their similarity, we suggest that an 1872-level event would be disruptive to modern electric power infrastructure.

The negative excursion from ≈18:30 to 19:05 UT and the positive recovery with overshoot from ≈19:05 to ≈20:38 UT can also be explained by the location of the observatory relative to the ionospheric Hall current, which is known as the DP2 equivalent current (Nishida 1968). The DP2 current flows in the counterclockwise direction on the duskside and clockwise on the dawnside when one views the Northern Hemisphere from the north pole. Since Greenwich was located on the duskside, it is likely that the positive ΔH excursions might correspond to the equatorward part of the enhanced DP2 current flowing in the eastward direction, while the negative excursion might correspond to the poleward part of it flowing in the westward direction. The DP2 current is a manifestation of the ionospheric convection. If the convection flow reversal was located near the open-closed boundary of the magnetic field lines (Milan et al. 2017 and references therein), then it is highly plausible that Greenwich was located in the polar cap during the interval from 18:30 to 19:05 UT, as the ΔH was negative (see also Valach et al. 2019).

The magnetogram from the Colaba Observatory in Bombay (N18°56', E72°50'; 100 MLAT), as shown in Figure 6, was preserved as a trace copy of the original observation (Basu 1954). There is much more structure in the H-trace at Greenwich for the C-S storm than for that recorded at Colaba. A common feature in both magnetograms is a sharp excursion in the H-trace at ≈18:30 UT. At Colaba, a strong negative excursion beginning near 18:00 UT resulted in a gap from 18:17 to 18:30 UT, during which a deflector magnet was inserted to keep the H-trace on scale. Following this gap, the H-trace exhibits a positive excursion of ≈300 nT within ≈20 minutes. During the time that the H-trace at Colaba was recovering, that at Greenwich was decreasing by a comparable amount.

We have also recovered a geomagnetic measurement at Tiflis (N41°43', E44°47'; 36.7° MLAT) from Russkoe Geografičeskoe Obščestvo (1873, pp. 6–10). This observatory recorded geomagnetic measurements in the eastward declination and in the horizontal force seemingly in a CGS unit. While the background H-value seems confused with that of the vertical force, the measurements' title and the actual values indicate the horizontal force. The Tiflis H-measurements are shown in Figure 7. These measurements are important because Tiflis is located approximately midway on a great circle line from Colaba to Greenwich. While the temporal evolution of the H-disturbance in the Tiflis magnetogram is similar to that in the Colaba magnetogram, the maximum amplitude in the H-component at Tiflis developed up to ≈1433 nT versus ≈1023 nT at Colaba (Basu 1954). The Tiflis H-measurements recorded the maximum at 15:01 UT (18:00 LT), a negative peak at 17:51 UT (20:50 LT), and a positive excursion up to 19:01 UT (22:00 LT).

Zoom In Zoom Out Reset image size

The median diurnal curve of the H-component in Tiflis (thick purple curve) varies within ±5 nT according to the February geomagnetic data in 1957–2001, and can therefore be neglected in the following analysis. Quiet time ΔH apparently lies within the range of one standard deviation (purple background in Figure 7), corresponding to the values between −30 and 15 nT. For the measurements at Tiflis, the lowest geomagnetic activity with an index equal to 30 nT was observed at 9 UT. Thus, the recorded 2436 nT H-value at 9:01 UT can be considered to be an undisturbed field. The uncertainty of this value can be estimated as ± the width of the quiet time deviations ranges, i.e., roughly ±50 nT. We derive the positive H excursion around 15 UT to be ≈219 ± 50 nT and the negative excursion around 18 UT as ≈1214 ± 50 nT, taking the local diurnal variation into account.

This storm was also recorded at Havana Observatory (N23°08', W082°21', 341 MLAT; Secchi 1872b). Havana sensed strong magnetic disturbances beginning at 10 LT (15:29 UT) and its magnetometer in the horizontal component went off-scale during ≈11:30–14:30 LT (16:59–19:59 UT; Secchi 1872b, p. 36). These magnetograms at Bombay and Havana (as described by Benito Viñes in Secchi 1872b) are consistent with each other in terms of temporal variation.

At Hanover VT (N43°42', W072°17', 551 MLAT), Twining (1872, pp. 279–280) described magnetic variations in the northeastern US between 10 and 14 LT (14:49–18:49 UT): "there were three epochs of extreme and sudden deviation followed by as many of sudden change back again,—that the extreme fluctuation was 5° 40' (in the declination) in three hours time, and that the violent disturbance preceded the visible (aurora)." The author also mentions that observations were suspended for 78 min before the local noon (15:31 UT).

The horizontal component, ΔH, of the Bombay magnetogram in Basu (1954, p. 50) shows annotated amplitudes of the SC of 85 nT (measured at 15:03 UT from the relatively stable pre-SC trace) and an H trace maximum-to-minimum (max-min) amplitude of 1023 nT (see also Moos 1910, p. 452), measured from the peak of the SC to the first measurement after the data gap from 18:17 to 18:30 UT, for which the value was chosen to give a max-min value of 1023 nT (assuming no further negative excursion during the data gap). Subtracting the SC amplitude from the maximal amplitude, the negative excursion in ΔH is estimated to be ≈−938 nT for the spot value (see Hayakawa et al. 2018). However, the annotated value of 1023 nT on which this value is based is likely to be a conservative estimate, unless the first reading from the deflected magnet was taken at the exact minimum in the storm H-trace.

Note that the ΔH value of ≈−938 nT is an instantaneous measurement obtained from the annotated value, whereas the Dst value should be computed by the hourly average of four mid- and low-latitude stations (Sugiura 1964; Sugiura & Kamei 1991; Akasofu & Kamide 2005; Siscoe et al. 2006; Gonzalez et al. 2011). We have further followed Dst calculation procedures to quantify the storm magnitude and time series in the Dst estimate (Dst*). These procedures require us to subtract baseline and background solar quiet (Sq) field variation from the hourly H variations for four stations and average them with latitudinal weighting (Sugiura 1964). The output value is called Dist H. We need Dist H from four mid/low-latitude stations to derive the Dst index (see Sugiura 1964). For now, we approximate the Dst estimate (Dst*) with Dist H, as it is extremely difficult to locate four magnetograms from mid/low-latitude regions for this storm.

Here, we have approximated the baseline with the pre-storm level and the Sq variation with the monthly H diurnal variation in Moos (1910, p. 64). On this basis, we have estimated the peak Dist H for the SC as ≈72 nT and the minimum Dist H for the storm as ≤ −834 nT, respectively. Accordingly, we approximate the min Dst* of ≤ −834 nT, as this value is likely to be an underestimate of the absolute value of the minimum H-trace excursion as noted above.

We have also computed the Dist H at Tiflis. Following Figure 7, the running hourly average value around 18 UT reaches the lowest value of 1312 nT. The diurnal H variation ranges ± 50 nT, following sigma bounds of the H-component at Tiflis according to the February geomagnetic data in 1957–2001. Taking into account the ΔH value of −1124 nT (1312 nT minus the background field of 2436 nT) and the magnetic latitude of Tiflis as 367, we estimate the lowest hourly Dist H at Tiflis ≈−1402 ± 62 nT, which is larger than the Dist H at Colaba, following the Dist H calculation procedure (Sugiura 1964). This can be reasonably explained by the asymmetry of the storm-time ring current (Clauer & McPherron 1980; Ohtani et al. 2007), which peaks in the dusk-midnight sector. It is not clear if the positive H excursion of ≈219 ±50 nT is entirely due to the sudden storm commencement and further investigation is needed to determine the origin of the positive H excursion.

These calculations allow us to conservatively estimate the minimal Dst* for this storm as ≤ −834 nT based on the Colaba magnetogram because the aurora was overhead at Tiflis and may have affected the local measurement (Russkoe Geografičeskoe Obščestvo 1873, pp. 6–10). In the following section, we reconstruct the equatorward boundary of the overhead aurora in the C-S storm down to ∣242∣ ILAT. From a relationship between the equatorward boundary of the overhead aurora and the minimum Dst value (Akasofu & Chapman 1963; Yokoyama et al. 1998; Hayakawa 2020; Cliver et al. 2022a) for great storms, Cliver et al. (2022a their Figure 39; see also Figure 9.4 in Hayakawa 2020) obtained a minimum Dst value of –1250 nT for 1872 storm. For now, we follow the conservative estimate for the minimum Dst* (≤ −834 nT) based on the Colaba magnetogram for the C-S storm.

This great magnetic disturbance was accompanied by significant auroral displays down to extremely low magnetic latitudes. We have investigated reports for auroral observations on 1872 February 4–6, following three contemporary auroral compilations for this storm (Preece 1872; Fron 1872; Donati 1874); auroral compilations from newspapers, diaries, and chronicles in South Asia and East Asia (Chapman 1957b; Hayakawa et al. 2018); the Yearbooks of the Russian Central Observatory (Wild 1874) and the American Chief Signal Office (ARSCO 1873); contemporary major scientific magazines [Nature (Nature 1872a, 1872b; Nature 1873), Meteorological Magazine (MMA 1872), Quarterly Journal of the Royal Meteorological Society (Royal Meteorological Society, 1873), American Journal of Science (Dana & Silliman 1872a, 1872b), Comptes Rendus (Académie des Sciences 1872), Wochenschrift zür Astronomie, Geographie, und Meteorologie (Heis 1873, 1874, 1876), and Zeitschrift der Österreichische Gesselschaft der Meteorologie (Jerinek & Hann 1872)]; newspapers in India, Australia, Iberian Peninsula, Azores, Bermuda, and the USA; several ship logs; and other miscellaneous reports (Tristram 1873; Capron 1879; Rixo 1994, p. 510; Al-Kurd al-Makk 2000, p. 370).

Figure 8 shows the distribution of auroral observation sites on 1872 February 4–6 based on the contemporary reports of visible aurorae. Aurorae were credibly reported for several locations below ≈20° MLAT, extending down to magnetic latitudes such as Bombay (100 MLAT) and Khartoum (134 MLAT), which are lower than the equatorward extent of visibility reported for either the Carrington storm in 1859 (∣173∣; Hayakawa et al. 2020) or the 1921 May storm (∣162∣; Angenheister & Westland 1921; Silverman & Cliver 2001). We can reconstruct the equatorward boundary of the auroral oval during the Chapman–Silverman storm based on the elevation angle of the auroral display and the magnetic latitude of the observational sites by assuming that the auroral emissions extended up to 400 km along the magnetic field lines (Roach et al. 1960; Silverman 1998; Ebihara et al. 2017). Accordingly, for events with elevation-angle reports, we can compute the lowest ILAT of overhead aurora, which is identical along the magnetic field line from the magnetic footprint (O'Brien et al. 1962), as shown in Figure 9, rather than the lowest MLAT of auroral visibility, which is the lowest latitude at which the aurora was actually observed. We selected the seven most credible low-latitude records for which an ILAT could be estimated (based on the detail/precision of the auroral reports) and summarize their details in Table 1. From these auroral records, the equatorward boundary of the auroral oval is considered to extend down to 24°–32° ILAT across the East Asian, Indian, African/European and eastern American sectors, i.e., ≈E140° to ≈W060°.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 1. Locations for the Most-equatorward Credible Observations of Aurora for the 1872 Storm from which an ILAT Could Be Determined

Site	Latitude	Longitude	MLAT	Max. in UT a	Elevation	ILAT
Shanghai	N31°14'	E121°29'	199	17:54	Zenith	242
Jacobabad	N28°17'	E068°26'	199	18:56	Zenith	242
Mauritius	S20°10'	E057°31'	−263	19:25	72°	305
St. Denis	S20°52'	E053°10'	−263	18:27–19:27	Zenith	296
Ispahan	N32°39'	E051°40'	266	N/A	60°	316
Sifook	S22°16'	E035°21'	−243	N/A	Zenith	279
Courland Bay	N11°13'	W060°47'	226	N/A	Halfway	292

Note.

^a This time can be either for sudden onset of the aurora, rapid change, or maximum sky coverage (i.e., climax of the event).

Download table as:  ASCIITypeset image

In the East Asian sector, the auroral displays were widely reported in Japan, Korea, and China; as exemplified in Figure 10 (Hayakawa et al. 2018). Among them, Mr. Vignale, General Consul of Italy in Shanghai, reported that "the bright arc extended almost for 50° very close to the zenith" at Shanghai (Donati 1874, p. 8). Considering the magnetic coordinates (N31°14', E121°29', 199 MLAT), we obtain an ILAT of ≈ 242 ILAT for this location (Hayakawa et al. 2018).

Zoom In Zoom Out Reset image size

In the Indian subcontinent, auroral emissions were reported in conjunction with telegraph disturbances between Jacobabad (N28°17', E068°26', 199 MLAT) and Lahore (N31°35', E074°19', 224 MLAT). At Jacobabad, the auroral display was "shooting from the east horizon to the zenith and very nearly at right angles to the magnetic meridian" and "after an hour there appeared in the aurora arch a little below the zenith a bright blue light of a dome shape quite intense; after some 15 minutes it suddenly dissolved to a deep violet" (Times of India, 1872-02-15, p. 2; Chapman 1957b, p. 187). Therefore, in this sector, we regard the reported lights not as stable auroral red (SAR) arcs (see Kozyra et al. 1997) with a monochromatic reddish appearance but as auroral displays themselves that extended down to 242 ILAT, and which were visible at Jacobabad (199 MLAT) over the zenith.

Heading westward, several low-latitude aurorae were reported around the Indian Ocean. In the southern hemisphere, aurorae were reported up to 72° in elevation at Mauritius (S20°10', E057°31', −263 MLAT) and up to zenith at St. Denis (CR65: S20°52' E053°10', −263 MLAT). On this basis, the auroral oval was reconstructed down to 305 ILAT and 296 ILAT, respectively. At St. Denis, "the colored part [of the aurora] to the south rose to the zenith" and "immense luminous jets [rose] from the horizon to the zenith, like columns of fire, bands, or flares of a whiter light" with significant brightness (Vinson 1872, p. 721). Here also, the aurora does not fit the description of an SAR arc. In the Middle East, several aurorae were reported in Iran (Preece 1872). At Ispahan (MI17: current Esfahān, N32°39', E051°40', 266 MLAT), the "whole of the Northern sky was crimson up to about 60°, like a magnificent sunset" (Preece 1872, p. 111). This indicates the equatorward extension of the auroral oval down to 316 ILAT, whereas this might alternatively indicate an equatorward boundary of the SAR arc based on the reported colouration and morphology.

On the eastern coast of Africa, the aurora was reported down to Sifook (S22°16', E035°21'; −243 MLAT) in current Mozambique, where the aurora was "almost extended to the zenith itself at one time in a broad belt" (Erskine 1875, p. 91). Considering the MLAT of Sifook and the reported elevation angle, we consider the auroral oval to have extended down to 279 ILAT.

In the American sector, the aurora was reported down to Tobago: "Mr. Taylor, master of the barque Tobago was riding at anchor at Courland Bay, and was a witness of this aurora. He described it ... as of 'a dark-red colour, extending half way up to the zenith and very brilliant, its situation being about NW by W." We interpret this report as an auroral display or an SAR arc extending 45° in elevation at Courland Bay of Tobago (N11°13', W060°47', 226 MLAT; Yeates 1872) off the northern coast of Venezuela; the auroral emission region is considered to have extended down to 292 ILAT.

Here, the reports at Esfāhan and Courland Bay describe monochromatically reddish glows. These glows may be SAR arcs given that only the red color is reported. SAR arcs have a dominant red color and are formless features of long duration (Cornwall et al. 1970, 1971; Kozyra et al. 1997). However, Jacobabad witnessed both the dynamics ("immense luminous jets [rose] from the horizon to the zenith") and coloration (as "a bright blue light of a dome shape") as reported in the Times of India (1872-02-15, p. 2; see also Chapman 1957b, p. 187) of the C-S aurora reported from other locations defy explanation in terms of SAR arcs. The aurorae at Sifook, Mauritius and St. Denis were also reported with whitish and purplish structures (Erskine 1875, p. 91; N25: Meldrum 1872a, p. 392; Vinson 1872, p. 721). Therefore, we consider that the auroral oval extended down to 24°–30° ILAT in its maximal intensity in each geographical sector.

Figure 11 shows the auroral visibility on 1872 February 4–5 as a function of magnetic latitude and time, with sites of auroral visibility ranging from 10° to 68°. The aurora was first noted by several observers near the SC time of ≈14:27 UT (as was the case for the 1989 March storm; Boteler 2019) or even earlier. For three stations (one in Korea and two in Japan), the reported auroral onset was ≈12 UT. Given the lack of geomagnetic activity preceding the SC, we suspect that these time stamps were erroneous because of lack of precise timing in East Asian records from the regular citizens during this period.

Zoom In Zoom Out Reset image size

The bulk of the plotted observers witnessed aurora at ≈20 UT and observations from Europe mostly ended by 24 UT, with only a few stations reporting aurorae after 6 UT on February 5. Note that the equatorward boundary of auroral visibility is not necessarily the same as that of the auroral oval because the aurora can be seen above the horizon, looking in most cases (see below) in the direction of the magnetic pole, given the auroral altitudes of ≈400 km. In particular, Figure 11(d) shows auroral visibility with non-reddish color around the negative Dist H peaks at ≈18 UT in Colaba and Tiflis. This concentration is interesting because these reports indicate that they are not SAR arcs but the robust aurorae on the basis of the inclusion of non-reddish colorations (see, e.g., Kozyra et al. 1997).

Here we consider the timings of dramatic onset of the aurora (e.g., Jacobabad), rapid change (Barnstaple), or maximum sky coverage (auroral climax) for comparison with magnetic records from Havana, Greenwich, Tiflis, and Colaba, with a focus on the low-latitude auroral observations from Table 1. Observers in East Asia witnessed aurorae during the earliest phase of this storm. The majority reported auroral onset around local midnight (14–15 UT) or a little earlier, with a​​​​​ ±1 h uncertainty in their local time system (Uchida 1992). The description of the aurora from Shanghai is from Von Gumpach: ²⁶ "The aurora was observed in Shanghai as a far-away explosion, of rather weak intensity. The sky at the horizon appeared dark, and the glowing arc covered almost 50°, near the zenith. We do not have an exact time for the beginning of the phenomenon. Its maximum was reached at around 2 am [17:54 UT] on February 5. The aurora stopped at 2:30 [18:24 UT], or shortly thereafter" (Donati 1874, p. 8).

In the Indian Subcontinent, the following account of the aurora from Jacobabad is from a correspondent to the Times of India (1872-02-15, p. 2): "As I was returning home about half-past 11 p.m. [18:56 UT], a sudden change from darkness to light was noticed as bright as the full moon. I was amazed; the conversion of Saint Paul came vividly before me—in fact, I was terrified by the sudden-change; my dog became motionless and seemed, to tremble. I thought it must have been a fire, no, the whole place was magically illuminated. ... After an hour there appeared in the Aurora arch a little below the zenith a bright blue light of a dome shape quite intense; after some 15 minutes it suddenly dissolved to a deep violet. The sky was clear and the phenomena was observed very distinct; it traveled towards the south and lasted till 2–15 (P.M.) [mistaken from A.M.: 21:41 UT] this morning, a duration of nearly 2 3/4 hours." The sudden onset of the aurora at 18:56 UT in Jacobabad occurred near the ≈19:01 UT end of the recovery (with overshoot) of a lower-limit (because of a data gap) > 150 nT negative excursion in the H trace at Colaba (Figure 6) that necessitated the insertion of the deflector magnet during the data gap from 18:17 to 18:30 UT, and near the 19:03 UT minimum of the Greenwich H-trace and onset of the data gap in the corresponding Z magnetogram (Figure 5). The H-trace behavior at Colaba (Figure 6) was mimicked by that in the Tiflis magnetogram (Figure 7), which recorded a negative H peak at 17:51 UT, and subsequent positive excursions from 17:51 to 18:01 UT of 221 nT and from 18:21 to 19:01 UT of 449 nT.

Observers in the southern Indian Ocean witnessed a great extension of the aurora australis. From the ship Pendragon (S13°43', E084°13'): "At midnight [≈18:23 UT] very suspicious-looking weather to the S., the sky being quite red" (Meldrum 1872b, p. 29).

The aurora as observed from Mauritius (S20°10', E057°31') by Meldrum (1872b): "9.48 p.m. [17:58 UT]—An irregular convex arch of dark red light extending over about 60° of the horizon, and having its vertex in the line of the magnetic meridian. ... 11 p.m. [19:10 UT]—Sixteen luminous bands of a steel gray to a silver white color, extending from as low down as I can see to within 20° of the zenith. ... 11.6 p.m. [19:16 UT]—The parallel bands are still seen. They cover the greater part of the hemisphere, extending (at the meridian) to about 72° above the horizon. ... 11.15 P.M. [19:25 UT]—A deep red glow from E. to W. by S. along the horizon. Fourteen parallel bands of a silvery color, with dark bands between them. They lie south and north, occupying nearly the whole southern hemisphere as far as the eye can reach, and are flanked at east and west by patches of blood and cherry red. ... 1.20 A. M. [21:30]—A bright red glow from S.E. to S.W. Intensest [sic] below the Centaur. Soon died away."

The aurora as observed at St. Denis (Réunion; S20°53', E055°27') by Vaillant (1872): "This was a magnificent aurora australis like I had never seen before. The first glows appeared around 8:30 [16:57 UT] ... its greatest intensity was reached between 11:30 in the evening [19:57 UT] and 1:30 in the morning [21:57 UT]." From the report by Vinson (1872): "This beautiful aerial phenomenon started around 8:30 in the evening [16:57 UT] ... From 10 to 11 [18:27–19:27 UT], the aurora seemed to reach the maximum of its intensity."

In the European sector, the auroral display was already visible after ≈16–17 UT (Earwaker 1872) from the southern horizon, and expanded to the northern sky with developments of distinct corona and rays (Figure 12; Earwaker 1872, pp. 322–324; Capron 1879, pp. 19–21). Townshend M. Hall (1872), a fellow of the Royal Geological Society, reported a "southern aurora" as "6. 0 p.m. [18:16 UT] Diffused red light near Orion's belt, in S.E. 6.15 [6:31 UT]" and "Several white rays in the South, diverging from near the Pleiades" at Barnstaple (N51°05', W004°04', 550 MLAT): "[up] to 6.55 [19:11 UT], no auroral light was distinguishable in the northern portion of the sky; but at that moment there was a sudden outburst of rays from the central point, covering the entire heaven in every quarter, several of the rays in the E. and E.N.E. being, however, especially remarkable for their width and color. The sky remained absolutely without a cloud until nearly 8 o'clock [20:16 UT], when there were only the remains of a few white streaks visible, and an hour later rain clouds, with a thick mist, obscured even the brightest of the stars" (Hall 1872, p. 2). The rapid expansion of the aurora to the northern sky at 19:11 UT at Barnstaple occurred within minutes of the lowest excursions in the Greenwich H- and Z-traces and near the peak of the positive bay following the 18:30–18:47 UT data gap in the Colaba H-trace, and falls near the sudden onset of the bright auroral display at Jacobabad at 18:56 UT.

Zoom In Zoom Out Reset image size

The aurorae were visible in the American sector only after the end of the storm's main phase. US weather maps for the local afternoon and evening of 1872 February 4 (16:35 LT and 23:35 LT in Washington, DC) show that the reporting stations for the US mountain states were under mostly cloudy skies, with snow obscuring the sky in several locations. ²⁷ A winter storm covered most of the inter-mountain west. The 23:35 LT synoptic (Washington, DC) discussion notes that aurora had been reported in New York state and at the Mobile, Alabama station. A newspaper report also noted a "splendid" overhead aurora observed in Austin, Texas (The Democratic Statesman, 1872-02-06, p. 3). In sum, aurorae were reported in the US, weather permitting, and the lack of auroral reports in much of the western US (See Figure 8) was likely to have been weather-related. Auroral visibility was reported down to the Courland Bay in Tobago (N11°11', W060°44', 225 MLAT), where it was witnessed at 19:00–21:30 LT (23:03–25:33 UT; Yeates 1872). This is consistent with a low-latitude auroral observation during the recovery phase of the storm, just prior to local sunset at Courland Bay (sunset ≈ 22:05 UT).

These comparisons of temporal auroral evolution at low magnetic latitudes and recorded magnetic activity indicate a close timing correspondence between key features in the evolution of both phenomena for the 1872 aurora. In particular, we note that the sudden appearance at 18:56 UT of aurora at Jacobabad (MLAT=199) occurred during the H-trace recovery from 18:31 UT—following a 14 minute gap during which a deflector magnetic was inserted at Colaba (100 MLAT) to keep the H-trace on scale during a sharp negative excursion—to a local peak at ≈19:01 UT. Corresponding behavior was observed in England (approximately five hours in local time and 40° distant in MLAT) where the aurora as viewed from Barnstaple (MLAT=550 N) suddenly filled the entire sky at 19:11 UT, near the minimum in the Greenwich (MLAT=550 N) H-trace at 19:03 UT. This is consistent with the Greenwich magnetogram that plausibly indicated Greenwich being located in the polar cap during the interval from 18:30 to 19:05 UT (Section 3.1).

In the 1872 February 6 issue of the Times of India, the aurora at Bombay was described as follows: "After sunset ... the aurora was slightly visible ... [and] It was distinctly visible until sunrise."

This report of a long-lived aurora at 100 MLAT was viewed with skepticism by Silverman (1995, 2006), although, as noted above, Silverman (2008) subsequently reckoned that the report from Bombay was credible. Silverman's initial difficulty in accepting the report of aurora in Bombay was based on the likely conflation of telegraph outages with a reported aurora at Singapore (−100 MLAT) for the great storm in 1909 September (Silverman 1995; Hayakawa et al. 2019a). The ≈10° gap in corroborating reports from sites between Bombay and Jacobabad may have been an additional factor. Our current investigation, however, indicates that is no longer the case for the Bombay report. Table 2 includes five stations between Bombay and Jacobabad on the ≈ W075° meridian, for which aurorae were reported with corroborating details. The MLATs for these five stations are well distributed between 129 MLAT and 176 MLAT (Table 2). While the timing of the aurora from sunrise to sunset at Bombay is problematic given the discordant detailed report from Jacobabad, the likely source of the report of aurora at Bombay (and also Aden; see Section 5.2) in the Times of India article appears to be beyond reproach.

George Benham Stacey (1839–1904) was one of the few people in Bombay on 1872 February 4 who would have had both the knowledge and motivation to look for an aurora in conjunction with the onset of the main phase of the great magnetic storm. He had a remarkable ≈ 50 yr career in the telegraph service, which began at age 12 when he joined the Electric Telegraph Co. His obituary (Anonymous 1904) details his rapid ascent up the ranks of his profession with postings at the Hague, Berlin, London, and Egypt, before a final appointment as the manager of the Eastern Telegraph Co. station in Bombay in 1872, where he remained for 28 yr. Quoting from his obituary, "Mr. Stacey was an ideal representative of the great cable service ... and no man in the service has ever been more universally respected alike by the staff and the directors of the companies in which he was engaged." In Preece (1872), Stacey's report from Bombay is succinct: "Strong earth currents 7.30 p.m. on 4th February. Continued until 7.0 a.m. on 5th. Aurora plainly visible 8:30 p.m. until 4:30 a.m." This report reduces some of the timing discrepancies in the auroral reports between Bombay and Jacobabad and the "plainly visible" repeats the text in the opening sentence of the February 6 Times of India article that appears to be an amalgamation from multiple sources (see Appendix). This report is also significant, reporting "strong earth currents" in Bombay (10.0° MLAT) immediately after the storm peak.

In sum, the evidence of corroborating reports between Bombay and Jacobabad, and the tribute to Stacey, combined with triangulation analyses by Chapman (1957b) and Hayakawa et al. (2018) that indicated that the overhead aurora inferred for Jacobabad could have been observed from Bombay at an elevation angle of 10°–15°, and new evidence for additional low-latitude aurorae visible at other longitudes as discussed in the following section, tip the scales in favor of the reality of the observation of aurora from Bombay in the C-S event.

The Times of India article from 1872 February 6 on the aurora ends with the sentence, "At Aden [83 MLAT] the aurora was brilliant in the extreme." Fron (1872) writes, "[The Aurora] was seen in Cairo, Egypt [278 MLAT], in Chartum [Khartoum; 134 MLAT], in Upper Egypt and probably—according to Mr. Raoul Pictet ²⁸ —up to Gondokoro ²⁹ [31 MLAT]". Fron (1872) annotated the tabulated Gondokoro observation with a question mark (?). Silverman (2008) could not uncover Pictet's original report and accordingly doubted the reliability of the records of aurora at these three stations, owing to the lack of specific details.

Our archival investigations have successfully located Pictet's original report in Archives des Sciences Physiques et Naturelles(Pictet 1872), where he reported telegraph disturbances and enhanced geoelectricity between Cairo and Khartoum in the local afternoon on 1872 February 4, Khartoum's inquiry to Cairo about the witness of a great reddish light on the horizon in that night, and his intention to search for further possible auroral visibility in Gondokoro. His report translates as follows: "I am coming back from the telegraph (office) where I collected the following information about the electrical phenomena that occurred yesterday evening. Yesterday in the afternoon, it was difficult to communicate with Khartum, which is located 15° North, 35° of East longitude. The [communication] devices were 'talking' by themselves: signs that were sent out were missing. The ground currents strongly impeded the service and the employees were fully confused, as they could not figure out what was causing those disorders. Yesterday evening, the office in Cairo got a dispatch asking what was the big red glow that was seen at the horizon, and suspecting a big fire. The telegraph line does not continue further South, beyond Khartum, but it is probable that this aurora has also be seen up to Gondokoro, at 5° North latitude. I shall inquire about it and I will confirm to you if they did see the aurora" (Pictet 1872, p. 74).

According to this report, the telegraph disturbance between Khartoum and Cairo is in at least rough temporal alignment with the climax of the geomagnetic storm in Figures 5–7. The auroral report from Khartoum looks realistic given that Khartoum confirmed the witness of great reddish light and inquired about its nature to Cairo. In contrast, Pictet only speculated a possibility of further auroral visibility down to Gondokoro without confirmation. This is probably why Fron assigned a question mark to Gondokoro in his list. Without independent confirmation, this is no more than Pictet's speculation. From Gondokoro, the auroral display could have been visible only up to 3° in altitude by triangulation, assuming the equatoward boundary of the auroral oval as ≈ 242 ILAT in this meridian as well.

On this basis, we confirm the auroral visibility down to Bombay (100 MLAT) and favor auroral witness down to Khartoum (134 MLAT). The brief report regarding Aden (83 MLAT) lacks details but was attributed to Stacey in Preece (1872). It is reasonable to assume that Preece was in communication with a local witness at Aden. Like the report of aurora at Singapore in 1909, the Aden report could be based on conflation with the disruption of telegraph communications because it was the first station after Bombay on the east–west portion of the link to London (Supplement to the Electrician 1894, p. 11). Based on triangulation, the aurora would be visible at these two sites (Aden and Khartoum) according to the equatorward boundary of the auroral oval. If we expect the equatorward boundary of the auroral oval to be ≈242 ILAT based on the reports from Shanghai and Jacobabad, then the calculated elevation angles at Aden and Khartoum would be 109 and 250, respectively. A report from Mecca (N21°25', E39°49', 177 MLAT; Al-Kurd al-Makk 2000, p. 370; see also Bekli & Chadou 2020) reduces a large gap in magnetic latitude between Aden (N12°49', E45°02', 83 MLAT) and other observational sites in West Asia such as Syene (N24°05', E32°56', 216 MLAT), Suez (N29°57', E32°34', 274 MLAT) and Sebbeh (Tristram 1873: N31°19', E35°22', 282 MLAT). These details make the Aden report credible, in our opinion.