Atlas of Historical Eclipse Mapsassets.cambridge.org/97805212/67236/frontmatter/...978-0-521-26723-6 - Atlas of Historical Eclipse Maps: East Asia 1500 BC-AD 1900 F. R. Stephenson

Atlas of historical eclipse mapsEAST ASIA 1500 BC - AD 1900

www.cambridge.org© in this web service Cambridge University Press

Cambridge University Press978-0-521-26723-6 - Atlas of Historical Eclipse Maps: East Asia 1500 BC-AD 1900F. R. Stephenson and M.A.HouldenFrontmatterMore information

http://www.cambridge.org/9780521267236

http://www.cambridge.org


Translation of cover textPart of a page from the Astronomical Treatise ofthe Chiu-t'angshu ('Old History of the T'angDynasty'). Among the astronomical records ofvarious kinds from around AD 760 is thefollowing detailed account of a total solareclipse:

'2nd year (of the Shang-yuan reign period), 7thmonth, day kuei-wei, the first day of the month.The Sun was eclipsed; the great stars were allseen. The Astronomer Royal, Ch'u Tan,reported [to the Emperor], "On kuei-wei day theSun was eclipsed. 6 k'o after the hour of ch'en(8.12 a.m.) the eclipse began. It was total at 1 k'oafter the hour of szu (9.12 a.m.). The Sun wasfully seen 1 k'o before the hour of wu(10.48 a.m.). The eclipse was 4 degrees inChang (lunar mansion). This represents Chou(state). The Chin-ti says that when the Sun iseclipsed between (the hours) szu and wu, thisrepresents Chou. Now Chou is Ho-nan, which isoccupied by Shih Szu-ming and his rebels. Thel-szu-chan says that under a solar eclipse thekingdom will be ruined".'

The recorded date of this eclipse correspondsto AD 761 August 5 and it can be seen from thecomputer-drawn map that on this day there wasa total eclipse visible at Ch'ang-an, the capital ofChina at the time. The recorded times are onlyapproximately correct but they agree quite wellwith calculation.






Atlas of historical eclipse mapsEAST ASIA

1500 BC-AD 1900

F. R. STEPHENSONDepartment of Physics, University of Durham

M.A.HOULDENDepartment of Physics, University of Liverpool

The right of theUniversity of Cambridge

to print and sellall manner of books

was granted byHenry VIII in 1534.

The University has printedand published continuously

since 1584.

CAMBRIDGE UNIVERSITY PRESS

Cambridge

London New York NewRochelle

Melbourne Sydney






cambridge university press Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi, Tokyo, Mexico City

Cambridge University Press The Edinburgh Building, Cambridge CB2 8RU, UK

Published in the United States of America by Cambridge University Press, New York

www.cambridge.org Information on this title: www.cambridge.org/9780521267236

© Cambridge University Press 1986

This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press.

First published 1986

A catalogue record for this publication is available from the British Library

Library of Congress Cataloguing Card Number: 85-47841

isbn 978-0-521-26723-6 Hardback isbn 978-0-521-10694-8 Paperback

Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate. Information regarding prices, travel timetables, and other factual information given in this work is correct at the time of first printing but Cambridge University Press does not guarantee the accuracy of such information thereafter.






To Ellen and Penny











Contents

IntroductionReferences

Eclipse maps1500 BC-AD 1900

viiiXV

1






Introduction

Of the various kinds of astronomical phenom-ena recorded in history, eclipses have possiblythe greatest interest and importance inpresent day research. Not only do they haveconsiderable value in chronological studies,they have also contributed much to ourunderstanding of changes in the Earth's rateof rotation, a major geophysical problem.

The astronomers of China have bequeathedto us the longest and most substantial seriesof eclipse observations from anywhere in theworld. Before about 700 BC the extant recordsare very sporadic, but since that date roughlyone thousand solar eclipses and severalhundred lunar eclipses are preserved. Thedetailed histories of Korea and Japan, bothcultural satellites of China, do not commenceuntil much later - towards the end of the 1stmillennium AD. However, from then on thereare extensive lists of independent eclipseobservations from both countries, in additionto those from China. For reasons given below,the present work is concerned only with solareclipses and covers the period from 1500 BC(several centuries before the earliest knowneclipse records in China) down to the end oflast century.

We have restricted our attention to solareclipses for two main reasons - historical andpurely practical.

Although there are several extant referencesto lunar obscurations from the earliest his-torical period in China, the Shang Dynastyc. 1550 to 1045 BC (Keightley, 1978), at allsubsequent periods down to the Sui Dynasty(AD 589-616) there are virtually no furtherrecords of these events. During this longinterval, only solar eclipses were regarded assignificant astrological omens and thusworthwhile reporting.

In studying historical eclipses, maps arevery useful for representing solar eclipsessince the local circumstances vary consider-ably from place to place. On account of boththe lunar orbital motion and the terrestrialrotation, when the Moon passes in front ofthe Sun the lunar shadow sweeps across theEarth's surface in a few hours. The umbralshadow forms a narrow band, usually morethan 10000 km in length but seldom morethan 250 km wide. Within this band the eclipsemay be described as central since the Mooncovers the Sun more or less directly. For aconsiderable distance on either side of thiszone a partial eclipse is visible. On the contrary,in the case of a lunar obscuration, mappingserves little purpose for the proportion of theMoon's disc in shadow is virtually indepen-dent of the observer's location; at any instantthe appearance is much the same from theentire hemisphere of the Earth where theMoon is above the horizon. Brief tables suchas those published by Oppolzer (1887) arethus quite satisfactory for investigating mostrecords of these phenomena.

Central solar eclipses may be divided intothree categories; total, annular, and 'annular-total'. A total obscuration, in which the Mooncompletely covers the Sun for a few briefminutes, producing a deep darkness with theappearance of stars, is certainly the mostawe-inspiring. However, if the Moon is in themore distant part of its elliptical orbit when itpasses in front of the Sun, an annular or ringeclipse is formed instead; here the diminutionof light may be quite small. Very rarely, aneclipse may be annular near the sunrise andsunset position. Under such circumstances,the eclipse track is extremely narrow and overpart of its extent the Sun is momentarily






reduced to a broken ring of light. In this Atlas,all tracks of central eclipses crossing the FarEast between 1500 BC and AD 1900 aremapped in detail, making use of recent exten-sive studies of the lunar motion and changesin the Earth's rotation in the past.

1. Calendar datesAll eclipse dates in this work are expressed interms of either the Julian or Gregorian calen-dars, the transition date being AD 1582 Oct5/15. On the BC/AD system there is, of courseno year zero, 1 BC being immediately followedby AD 1. Hence in specifying years of eclipsesbefore the Christian Era we have used negativenumbers, equating the year 0 with 1 BC, - 1with 2 BC, etc. Thus the year of the earliesteclipse charted in this Atlas (-1496) corre-sponds to 1497 BC. This system is followedby Oppolzer (1887) and is in common use inastronomical practice. A number of otherdates such as reference epochs are given inthis introduction; for these we have adoptedthe BC/AD system since negative numberscould possibly be mistakenly interpreted asrelating to quantities other than dates.

The calendar date which we assign to anyparticular eclipse is the local date and is inde-pendent of the Greenwich meridian. In theregion covered by the maps, all places wherea particular eclipse is visible share the samecalendar date. This date will frequently beone day in advance of the Greenwich civildate (e.g. as listed by Oppolzer, 1887) sincethe selected range of longitude is far to theeast of Greenwich. Eclipse dates as givenhere should be identical with the correspond-ing historical date as converted to theWestern calendar, useful conversion tablesbeing those of Hsueh and Ou-yang (1956).

Following the calendar date of each eclipsewe have given the Julian day number at thetime of conjunction; this is rounded to thenearest integer. Julian ephemeris days, be-ginning at 12h ET (ephemeris time), are usedhere. The quoted day number thus may exceedby unity the appropriate value given byOppolzer (1887), which is based on UT(universal time).

2. Fluctuations in the Earth's rotation and theaccuracy of the eclipse mapsIf the Earth's rate of rotation and the meanlunar motion were uniform, it would be poss-ible to calculate with high accuracy the geo-graphical co-ordinates of the zones of totalityor annularity across the Earth's surface foreclipses at any time in the past (or future). Inpractice this is not the case, mainly on accountof the tides produced by the Moon and Sun.During the course of a century, tidal friction inthe oceans and - to a lesser extent - in thesolid body of the Earth causes a simultaneouslengthening of the day by some 2.5 milli-seconds (ms) and a recession of the Moonfrom the earth by some 3.5 metres. As theMoon's orbit expands in this way, its meanmotion slows down (in accordance withKepler's third law). The equivalent orbitalacceleration is approximately -25 arcsec/century2 (7cy2), producing a substantial driftin the lunar position over a period of a fewmillenia. A detailed discussion of the problemis given by Stephenson and Morrison (1984),to which the interested reader is referred.Unless otherwise indicated, the results givenin this section are taken from this paper.

In all calculations of the Moon's positionwe have used the lunar ephemeris j=2 (IAU,1968), but with a revised value for the lunaracceleration (h). We have adopted a figure forh of -26 7cy2, as deduced by Morrison andWard (1975) from transits of Mercury observedover the last 250 years. This result is verysimilar to that determined from lunar laserranging by Dickey and Williams (1982), i.e-25.1 ± 1.2. Over the period covered by thisAtlas it is unlikely that tidal friction haschanged significantly and we have thusassumed a constant h throughout. Theephemeris j=2 is based on a value for h of-22.44 7cy2. In order to fully incorporate ourchoice of h we have applied the followingadditions to the mean lunar longitude (L) ofthe ephemeris:

A L = - 1 ".54 + 2".33 T - 1 ".78 T2(1)

Here T is measured in centuries from theepoch 1900.0.

Allowance for variations in the rate of ro-tation of the Earth is more complex, since

IX






both tidal and non-tidal changes are present.On the basis of conservation of angularmomentum in the Earth-Moon system, therate of lengthening of the day due to thecombined effect of the lunar and solar tidesmay be calculated fairly accurately for a givenvalue of the lunar acceleration. With h = -267cy2, the result is 2.4 ms/cy. Although thismay seem very small, the resultant accumu-lated clock drift (AT) over the historical periodis large; in T Julian centuries (each of 36525days) it amounts to some 44 T2 seconds.Hence by the beginning of the period coveredby this Atlas (1500 BC) the tidal contributionto AT is roughly 0.5 day. Neglect of thiswould place eclipse tracks on the oppositeside of the Earth from where they wouldotherwise be. The importance of carefullyallowing for changes in the Earth's rotationwhen investigating past records of eclipses isthus obvious.

Both precise modern measurements andrelatively crude pre-telescope observations(mainly of occupations and eclipses) revealsignificant non-tidal fluctuations in the Earth'srotation which materially affect AT. Irregularshort-term changes, the so-called 'decadefluctuations', which can be accurately mappedover the last two centuries, are usually at-tributed to electromagnetic coupling betweenthe core and mantle of the Earth. These pro-duce deviations from the tidal AT parabolaby up to 30 seconds of time. Pre-telescopicobservations are much too inaccurate to revealthe decade fluctuations but they do provideclear evidence of significant non-tidal changeswhich are variable on the time-scale of mil-lennia. Possible causes are (i) long-termelectromagnetic coupling, (ii) redistributionof material in the interior of the Earth and(iii) global sea-level changes. Analysis ofBabylonian observations (mean epoch around400 BC) reveals that these non-tidal effectsreduce the expected (tidal) value of AT bymore than an hour in ancient times.

Above each map, beside the date, we havegiven the estimated value of AT, based on adetailed investigation of (i) telescopic timingsof occupations of stars by the Moon; (ii)medieval timings of lunar and solar eclipsesby Arab astronomers; (iii) medieval andancient sightings of total solar eclipses ob-

served mainly in Europe and China; and (iv)ancient timings of lunar eclipses by Babylonianastronomers. This analysis is based on thesame value for h as adopted here (i.e. - 2 67cy2). Our main source of reference is againStephenson and Morrison (1984). It should benoted that the pre-telescopic values of AT areactually calculated on the basis of a preliminarystudy, the results of which differ only slightlyfrom those of the main paper.

Between AD 1600 and 1900, we have de-duced AT values from the data at yearly and5-yearly intervals published by Stephensonand Morrison (1984). Before AD 1600, thefollowing parabolic expressions were usedinstead, AD 948 being the mean epoch of theArabian timings mentioned above.

(i) from AD 948 to 1600,AT = 22.512 sec (2)(ii) at any time before AD 948,AT = 1830 -405E +46.5E2 sec (3)

Here, t is measured in centuries from AD1850 and E in centuries from AD 948.

The value of AT exceeds 1 minute duringthe whole of the period before AD 1640. Overthis entire range of time we have rounded theindividual values to the nearest 0.1 minute;later we have gradually increased the quotedprecision. Some remarks are necessary onthe real accuracy with which AT can be de-termined in the past since this has importantbearing on the reliability of the maps andtables.

Reasonable estimates of the uncertainty inthe adopted values of AT at century intervalsback to AD 1600 are as follows: AD 1900, 0.1sec; 1800, 1 sec; 1700, 5 sec; 1600, 30 sec.Because of the sporadic nature of the pre-telescopic data it would be impractical toquote values at similar intervals before AD1600. However in the medieval period theuncertainty is not much more than a minuteand even well into ancient times the value ofAT is probably known to within a very fewminutes. Thus the groups of Arabian andBabylonian timings give results for AT withstandard errors of 1 min 20 seconds at AD 948and approximately 7 minutes at 390 BC. Poss-ibly a realistic estimate of the uncertainty inAT by 1500 BC is around 15 minutes. Obvi-ously these latter estimates are much larger






than the appropriate rounding errors (0.1 min).However, for purposes of reference we havefelt it advisable to indicate the precise valueof AT used in computing the data for eachmap.

With regard to the reliability of the eclipsemaps and tabular data, the following remarksseem especially relevant. Uncertainties in ATare directly reflected in the computed localtimes of maximum eclipse. However, as thelatter are rounded to the nearest minute, errorsonly become significant in the more ancientpast. In practice, as eclipse times are seldomrecorded in Chinese history before about AD1500 - and then only to the nearest doublehour - the calculated times may be regardedas of more than adequate precision. A detailedknowledge of the accuracy of the map tracksonly becomes important when an observationalleges a very large eclipse. However, suchrecords deserve special attention since theyare of value in refining AT. In calculatinggeographical co-ordinates, we have regardedlongitude as the independent variable,deducing the latitudes of the edges of thebelts of totality for discrete values of longitude.Uncertainties in AT produce errors in theselatitude limits at each selected longitude.These errors depend also on the angle whicha particular track makes with the equator atthe appropriate longitude. Thus if a track oftotality or annularity runs nearly parallel tothe equator at some point, even quite a largeuncertainty in AT will have negligible com-ponent in latitude. For other angles of incli-nation (I) to the equator, it is readily seen thatthe appropriate latitude error is approximatelyproportional to tan I. The Earth turns through0.25 deg in 1 minute of time. Hence if we let Mminutes be the estimated uncertainty in ATon a particular date, the corresponding errorin the latitude limits is roughly (M tan l)/4. Ona typical eclipse map, the mean value of I issome 25 deg (tangent close to 0.5). Thusaround the epochs AD 1600, AD 1000 and 500BC, likely latitude errors at any given longitudeare respectively 0.06, 0.15 and 0.8 deg. Eventhis last amount represents only about 2 mmon the scale of the maps so that the mappingaccuracy is expected to be high, even inancient times.

On the question of investigating hitherto

unused records of large solar eclipses fromthe Far East to deduce AT more accurately,possibly the best prospects would seem torelate to the more ancient observations; thematerial since about 200 BC has been exten-sively studied. At present, only three recordsof total solar eclipses are known before 200BC, all from the chronicle of a single smallstate - Lu, the home of Confucius. The datesof these events as recorded in the Ch'un-ch'iuare as follows: 709, 601 and 549 BC. If anyadditional observations from the Shang orChou Dynasties were to come to light theymight revolutionise present knowledge of thehistory of AT. It is hoped that the productionof this Atlas will encourage research in thisfield.

3. Capital cities of China, Korea and Japan

(i) ChinaOn each of the eclipse maps we have marked,for reference, the site of a major Chinese city- usually the capital of the time (the Wade-Giles system is used in the transliteration ofnames). Throughout much of the long historyof China, because of the great importanceattached to political astrology, the hub ofastronomical activity was the metropolis.Most of the solar eclipses (as well as othercelestial phenomena) recorded in the variousdynastic histories can thus be assumed tohave been observed at the appropriate capital.However, the earliest preserved series of solareclipse observations (from any part of theworld) probably does not come into thiscategory. During the so-called 'Spring andAutumn Period' (722 to 481 BC), more than 30solar eclipses are recorded in the Ch'un-ch'iu.It seems probable that most of the obser-vations were made at the state capital, for theCh'un-ch'iu notes several eclipse ceremoniestaking place there.

In later times, we find occasional statementsto the effect that a particular eclipse was notseen by the Astronomer Royal (e.g. on accountof cloudy weather at the capital) but wasreported from some provincial city instead.However, these appear to be only randomsightings. Once the empire was centralised(221 BC) there is no evidence of provincialobservers maintaining a systematic watch of

XI






the sky in the style of the astronomers at theimperial court.

The capital of China has been changed manytimes, any particular place seldom havingheld the honour for more than about twocenturies. Often a move took place as theresult of a new dynasty being established butnot infrequently the change occurred onaccount of the loss of the original capital dueto invasion or rebellion. At several periodsthe emperor and his court were in temporaryexile, while the repeated movements of theTang court back and forth between Ch'ang-an and Lo-yang over several decades aroundAD 700 are well known to historians. We haveonly taken into account what we regard as themore important changes of capital. Never-theless, there must be a degree of arbitrarinessabout our choice at certain critical times.

Several major Chinese cities have beenknown by different names at different periods.Thus, close to the site of the Former Han (202BC - AD 9) metropolis of Ch'ang-an, the firstSui emperor founded his capital of Ta-hsingCh'eng (AD 583). This was later (AD 618) re-named Ch'ang-an by the first Tang ruler. Ateach period we have given the style by whichthe capital was usually known at the time, butwe have avoided name changes during adynasty.

Apart from minor periods, the capital ofChina has occupied a site near one of thefollowing seven present-day cities: An-yang(pinyin spelling: Anyang), Hang-chou(Hangzhou), Hsi-an (Xian), K'ai-feng (Kaifeng),Lo-yang (Luoyang), Nan-ching (Nanjing) andPei-ching (Beijing). Table 1 lists in alphabeticalorder the various Chinese cities marked onthe maps together with their geographical co-ordinates (in degrees and decimals). Notethat the Tang city of Ch'ang-an was located alittle to the south-east of the Han city of thesame name.

During the earliest truly historical dynasty -the Shang (c. 1550 BC to 1045 BC) - thecapital appears to have changed several times.However, the details are obscure and thedates uncertain. Original Shang records - inthe form of 'oracle bones' - have, however,been recovered only from a single site, the'Wastes of Yin' near An-yang. The vastnumbers of inscribed bones excavated here

Table 1. Locations of Chinese cities markedon the various maps

Name of City

Ch'ang-an (Han)Ch'ang-an (Tang)Ch'eng-tuChien-k'angHaoHsien-yangHsu-changK'ai-fengLin-anLo-iLo-yangPei-chingTa-hsing Ch'engTa-tuYinYing-t'ien

Latitude

34.35°N34.2730.6232.0334.1634.3934.0534.7830.2534.7534.7539.9234.2739.9236.0732.03

Longitude

108.88°E108.90104.10118.78108.72108.85113.80114.33120.17112.50112.47116.42108.90116.42114.33118.78

(Keightley, 1978) indicate the importance ofthe place over a considerable period. It seemspossible that the capital which was traditionallycalled Yin occupied this site and that this isrecalled in the modern name for the locality.Lacking direct evidence to the contrary, wehave assumed Yin to have been the capitalduring the entire period from the beginningof this Atlas (1500 BC) to the fall of the ShangDynasty around 1045 BC (Pankenier, 1984).This is almost certainly an over-simplification,but any other Shang capitals were probablyin the vicinity of this place.

During the subsequent Chou Dynasty(c. 1045 to 256 BC - if we include the Chan-kuoor Warring States Period) the capital was firstat Hao, but for the whole of the period after770 BC it was located at or near Lo-i. Around516 BC the royal residence was transferred toCheng-chou, some 10 km to the east of Lo-i,where it remained for two centuries. However,we have felt it unnecessary to incorporatesuch a trivial change in position. For reference,the geographical co-ordinates of the capitalof the state of Lu (modern Chu-fu) wheremost of the extant observations between 722and 481 BC were probably made are: lat 35.13deg N; long 117.02 deg E.

Finally, we come to the question of theChinese capital at other major periods of par-

XII






tition, notably the Northern and SouthernDynasties (AD 420 - 589) and the co-existenceof the Kin and Sung Empires (AD 1127 -1234). In each case we have marked themetropolis of the native Chinese Dynasty(respectively Chien-k'ang and Lin-an) on theappropriate maps. However, it should beemphasised that there was also much inde-pendent astronomical activity centred largelyon the site of Ch'ang-an (AD 420 - 589) andK'ai-feng (AD 1127 - 1234). In consequence,there are many parallel reports of eclipsesand other astronomical phenomena at thesetimes.

(ii) Korea and JapanEclipse records from these countries coveronly about the last third of the time-span ofthis Atlas and in order to maintain a systematicformat for the maps we have not marked thesites of the appropriate capitals. Instead, wehave preferred to summarise the historicalbackground. Because of the small extent ofboth Korea and Japan compared with China,the spatial movements of the various capitalsare relatively mihor on the scale of the maps.

A significant number of eclipse records arefound in the Samguk Sagi, the earliest historyof Korea, coveflhg the period from the 1stcentury BC t6 AD 935. However, these recordsare generally acknowledged to be of dubiousreliability, some possibly having Originated inChina. The detailed history of Korea do£s notbegin until the establishment of the kingdomof Koryo in AD 918, although the first eclipserecord in the Koryo-sa is as late as AD 1012.However, from this time onwards, both solarand lunar eclipses are fairly regularly nbted inKorean history.

The capital of Korea was Songdo (lat 37.97deg N; long 126.58 deg E) for almost theentire Koryo Dynasty (AD 918 - 1392) andapart from two periods of exile, we shouldexpect the eclipse records to originate fromthere. It seems likely that eclipses charted inthe periods AD 1236 to 1268 and 1290 to 1292inclusive were observed on the island ofKangwha (37.73 deg N; 126.48 deg E), towhich the court had fled during periods ofMongol invasion. Soon after the fall of theKoryo Dynasty, the newly established YiDynasty moved the capital to Hanyang

(modern Soul, 37.55 deg N; 126.97 deg E) inAD 1394. Here it has remained apart frombrief periods during which no solar eclipsesare charted.

The first solar eclipse record from Japandates from AD 628 and from this time onwardseclipses are at first occasionally and later fairlyfrequently reported. However, the historicalsources are by no means as systematic asthose from China or Korea. Kanda (1935)compiled a valuable list from diaries of court-iers, temple records, etc as well as histories.

From very early times, the Japanese capitalmoved from one site to another in the YamatoPlain following the death of each emperor.Nara (34.68 deg N; 135.82 deg E) was estab-lished as the first fixed capital in AD 710, butthe city held this position only until AD 784.For a further decade, Nagaoka (37.45 deg N;138*83 deg E) became the imperial residence,after which the capital was established atHeian (modern Kyoto, 35.03 deg N; 135.75deg E). Here it remained for more than athousand years until it was replaced by Tokyo(35.67 deg N; 139.75 deg E) in 1868.

In AD 1185, a Shogunate was founded atKamakura (35.32 deg N; 139.55 deg E) andthis became an important cultural centre. It isthus likely that a number of eclipse obser-vations noted in Japanese history during thisperiod were made here instead of Heiah.Similar remarks apply to Edo (Tokyo, 35.67deg N; 139.75 deg E) between about AD 1590and 1868, after which it became the capital.

4. Data represented on the mapsAbove each map is given the following in-formation: Julian ephemeris day number(rounded to the nearest integer), local calehdardate (year, month and day) and value of ATused in making the various calculations. Onthe map itself is shown that portion of a trackof totality or annularity which lies within theselected area: from 15 to 45 deg N and from 90to 150 deg E. (The adopted sign convention forlongitude is negative to the E of Greenwich).The edges of the zones of totality are shownby solid lines, those of annularity by brokenlines. If an eclipse was annular for part of itstrajectory and total for the remaining part thetrack edges are shown by solid lines. Theactual phase at any particular longitude can

XIII






be determined from the data listed below themap. Where appropriate, sunrise and sunsetpositions are denoted by short lines a little tothe west or east of the main track and roughlyperpendicular to it.

Below each map is given the degree ofobscuration of the Sun at the E and W ends ofthe map track (over the range of longitudeslisted to the right of the map). For a totaleclipse, these figures will exceed 100 per cent,while for an annular eclipse the converse willbe true. Where appropriate, the followingadditional information is supplied: (i) thegreatest obscuration of the Sun along theeclipse track - if a maximum occurs in theregion covered by the map; (ii) the approxi-mate location at which the Sun rose or setcentrally eclipsed, allowing for refraction andsemi-diameter.

Accurate co-ordinates of an eclipse trackare given in the space to the right of eachmap. Over a range of longitudes, usually fromabout 87 to 153 deg E, are given the followingdetails: latitude of the N limit of the eclipsetrack, latitude of the S limit, approximate localtime (in hours and minutes) of greatest phaseand mean altitude of the Sun at this time. Thelatitude information will enable the user todraw the track on a more detailed map orinvestigate more carefully the local circum-stances at a particular place, if required.

5. Comparison with historical records(i) Total and near-total eclipsesSince all dates in this section directly concerneclipses, we shall use the convention ofpositive and negative numbers rather thanthe BC/AD system. Eclipses which were eithertotal or nearly so are reported at almost allperiods in Chinese history and there areoccasional records from Korea and Japan.Such records are particularly prevalent in theastronomical treatises of the Hanshu andHou-han-shu (which between them cover theperiod from about -200 to +220). It isinstructive to compare these observationswith the computed paths of totality andannularity as shown on the appropriatemaps. The recorded dates, as converted tothe Julian Calendar, and descriptions aregiven in Table 2.

Table 2. Han records of large solar eclipses

Julian Date Description

-197Aug7-187July17-180March4-146Nov10-88Sep29-79Sep20-33Aug23

-27Jun19-1Feb5+ 2Nov23+65 Dec 16+ 120Jan 18

totalalmost completetotalalmost completenot complete, like a hookalmost completenot complete, like a hook;it setnot complete, like a hooknot complete, like a hooktotaltotalalmost complete, daybecame like evening

From Table 2 it will be seen that four totalsolar eclipses are recorded (-197, -180, +2and +65) but without any details, although theShih-chi states that on the occasion of theeclipse in -180 it became dark in the daytime.Reference to the maps shows that all but theeclipse of +2 are represented as central (onlymarginally so in +65) at the capital of the time(Ch'ang-an or Lo-yang) but in the case of -197the phase was annular instead of total. Itwould seem that there was no separate termat this period to describe a ring eclipse; theannular obscuration of +516 is reported in thesame way (in the Nan-shih). The observationin +2 is clearly discordant for a large change inAT would be necessary to render the eclipsecentral at Ch'ang-an. In fact, since the belt oftotality was of virtually negligible width, theprobability of the central phase actually beingwitnessed and reported to the capital is min-ute. Possibly an original report of an 'almosttotal' eclipse was later abbreviated.

Of the eclipses described as either 'notcomplete, like a hook' or 'almost complete' therelevant maps show that in -187, -146, -88,-79, -27 and - 1 a large partial eclipse wasindeed visible at the capital. The map for+120makes this eclipse actually total at Lo-yang butonly a small error in AT (some 3 minutes)would render the phase just partial - inkeeping with the record, which alleges a verylarge obscuration. There was no eclipse visi-

XIV






ble in China in -33, but the fact that the textalso indicates that the Sun set while theeclipse was still visible led Dubs (1941) tosuggest an alternative date of -34 Nov 1. Onthis occasion the Sun would set eclipsed butthe dating error is difficult to explain; it will benoted that the other dates of the eclipsesdiscussed here agree exactly with the calcu-lated dates.

(ii) Partial eclipses

Although the main emphasis in this work is onthe accurate mapping of the paths of centraleclipses, most large partial eclipses visible inthe Far East should also be identifiable. Arough rule for estimating the magnitude of aneclipse outside the central zone is as follows.On the scale of the maps, a displacement of1 mm perpendicular to the edge of a track oftotality or annularity will decrease the degreeof obscuration of the Sun by some 1 per cent.In the case of a total eclipse, the calculatedmagnitude on the edges of the map track isexactly 100 per cent, so that interpolation isstraightforward. However, for an annular

obscuration, allowance must be made for thefact that even the central obscuration may bemuch less than 100 per cent (see details givenbelow map).

If an eclipse track passes a little to the N or Sof the region covered by the maps, there willbe no entry for this particular date despite thefact that a partial eclipse will be produced overmuch of our area of interest. Hence it isapparent that this Atlas is by no means com-plete for partial eclipses visible in the Far East.

As a practical illustration of the coverage ofobserved eclipses of any magnitude we maytake the case of the Ch'un-ch'iu recordalready referred to above. Between —710 and-480, this chronicle notes 37 eclipses, mostlywithout any descriptive details. Four of thesemust be presumed to be either false sightingsor possibly abortive predictions since on ornear the stated dates no eclipse was visible onthe Earth's surface. Of the remaining 33, asmany as 28 eclipses are charted here, theexceptions being of fairly small computedmagnitude in NE China. A similar degree ofcompleteness seems likely at other periods inFar Eastern history.

ReferencesDickey, J. O. & Williams, J. G. (1982).Geophysical applications of lunar laserranging. Trans. Amer. Geophys. Union, 163,301.Hsueh Chung-san & Ou-yang I (1956). ASino-Western Calendar for Two ThousandYears: 1-2000A.D. Peking, San-lien Shu-tien.IAU (1968). Trans. Int. Astr. Union, 13B, 48.Kanda Shigeru (1935). Nihon Tenmon Shiryo.Tokyo: Koseisha.Keightley, D. K. (1978). Sources ofShangHistory. Berkeley: University of CaliforniaPress.

Morrison, L V. & Ward, C. G. (1975). Ananalysis of the transits of Mercury:1677-1973. Mon. Not. Roy. Astr. Soc, 173,183-206.Oppolzer, T. R. von (1887). Canon derFinsternisse. Wien: Karl Gerold's Sohn.(Reprinted 1962 as Canon of Eclipses. NewYork: Dover Publications, Inc.)Pankenier, D. W. (1982). Astronomical datesin Shang and Western Zhou. Early China, 7,1-37.Stephenson, F. R. & Morrison, L V. (1984).Long-term changes in the rotation of theEarth: 700 B.C. to A.D. 1980. Phil. Trans. Roy.Soc. A, 313,47-70.

xv











Documents

Atlas of Historical Eclipse Mapsassets.cambridge.org/97805212/67236/frontmatter/...978-0-521-26723-6 - Atlas of Historical Eclipse Maps: East Asia 1500 BC-AD 1900 F. R. Stephenson