Lot 170
  • 170

A rare Umayyad brass astrolabe, signed by Muhammad ibn al-Saffar, Spain, Cordoba, dated in Western Abjad 411 AH/1020 AD, with later Ottoman Turkish rete, 16th/17th century

300,000 - 500,000 GBP
bidding is closed


  • brass
of typical round form, with simple, stylised throne, leading to 2 suspension loops, the reverse of the mater with central circular recessed section, original alidade with foliated terminals, later 16th/17th-century Ottoman rete with cursive inscriptions, replacement pin


Private collection, France.

Catalogue Note

This is the oldest known dated astrolabe from Muslim Spain - the mater and plates of a hitherto-unrecorded astrolabe made by the celebrated Andalusi astrolabist Muhammad ibn al-Saffar in Cordoba, dated 411 AH/1020-21 AD. It is a significant addition to the canon of known astrolabes from al-Andalus, and of major historical interest.  

This important new witness to the vibrant scientific tradition in Muslim Spain is the earliest known dated astrolabe from the Islamic East. It was made by a celebrated early-eleventh-century instrument-maker of Cordova already known by two other astrolabes. The rete is a late medieval (sixteenth/seventeenth century) replacement, from the lands of the Eastern Mediterranean. It fits perfectly, which suggests that the instrument had a second life far from its original habitat.

Astronomy in al-Andalus

Astronomy flourished in al-Andalus, a term used for that part of the Iberian Peninsula under Muslim domination at any time, for several centuries. The new Islamic astronomy from the Islamic East imposed itself on the traditional folk astronomy of Visigothic Spain. The resulting Andalusi school of astronomy was progressive and actively encouraged, but it lacked the stability and ingenuity of the main regional schools of the Islamic East: Iraq, Iran and Central Asia; and Egypt, Syria and the Yemen. Thus, for example, the major astronomical handbooks with tables and explanatory text that were used in al-Andalus were those of al-Khwarizmi of Baghdad (c.825 AD, based on Indian theories) and al-Battani of Raqqa (c.910 AD, based on Ptolemaic models). By the eleventh century these were out-dated in the Islamic East and had been replaced by superior works, but in al-Andalus they reigned supreme and formed the basis of the Latin Toledan Tables, a hotchpotch of tables compiled c.1080 AD that was very popular in Europe as far as England until the fifteenth century. Europe never knew the wealth of Islamic astronomy in either the East or the West until orientalists, mainly French and German, began to investigate it in the nineteenth century. In the twentieth century a succession of Spanish historians in Barcelona – primarily José Mìllas Vallicrosa, Juan Vernet and Julio Samsó, and their students – have documented the history of Andalusi astronomy, paying due attention to the folk tradition and to instrumentation (two subjects usually given short shrift by historians of astronomy).

One well-known Andalusi astronomer was Ahmad ibn al-Saffar (d. Denia, 1035 AD). He taught arithmetic, geometry and astronomy in Cordoba, and together with his teacher al-Majriti, prepared a recension of the astronomical handbook of al-Khwarizmi, of which the Arabic original is lost but we have a Latin translation (published by H. Suter with an English translation by Otto Neugebauer). He prepared a treatise on the use of the astrolabe written in a “clear, simple and comprehensible style’’ which was used in Europe in Latin translation until the fifteenth century (see Sheynin for the most recent study). Ahmad also made a marble sundial for the latitude of Cordoba which survives in the Museo Arqueológico in that city (see King, ‘Sundials from Andalucia’, and article ‘Mizwala’ in Enc. Islam, 2nd edn.). But it is Ahmad’s brother Muhammad who concerns us here.

The celebrated Muhammad ibn al-Saffar and his three known astrolabes

The historian Sa’id al-Andalusi wrote in his universal history that Muhammad ibn al-Saffar constructed astrolabes like no one before him (Tabaqât al-umam, text, p.70, transl., p.131). Two other astrolabes signed by this maker have been known for some time, one for over half a century, the other for over a century and a half. They enable us to confirm the authenticity of this new astrolabe, for the distinctive engraving on all three pieces is identical. Yet there are subtle differences in the details.

The astrolabe by Ibn al-Saffar dated Cordoba, 417 AH/1026-27 AD, preserved in the Royal Scottish Museums (inv. no. T1959-62), Edinburgh, is missing the rete, a replacement being of Eastern Islamic provenance. It has never been properly published (which means more than publishing a picture of the front).

The astrolabe dated Toledo, 420 AH/1029 AD, is preserved in the Deutsche Staatsbibliothek (Orientabteilung 2050), Berlin, and is complete. The plates for Cordoba and Toledo have the names of these cities added in a later hand. The piece was published in a model fashion by the orientalist Franz Woepcke in 1858 (see also Mayer, Islamic astrolabists, p.75). A single plate, preserved in the Museo Nazionale, Palermo (4025), can be identified as the handiwork of Ibn al-Saffar by means of the maker’s distinctive engraving. Its bears markings for latitude 40°, Toledo, on one side, and for latitude 42;30° (locality unspecified) on the other.

We can be confident that the present astrolabe by Ibn al-Saffar was originally graced with an elegant but delicate rete such as survives on the Berlin astrolabe. Indeed, his retes were too delicate and fragile for practical purposes, not least because the star-pointers tended to be overly long. Only one out of three has survived a millennium.

The earliest astrolabes from al-Andalus number less than twenty and are well documented (that is, detailed descriptions are available), if not yet published. Specialists have been surprised by the appearance of another such astrolabe, hitherto unknown to scholarship.

An illustration of a tenth-century Andalusi astrolabe signed by Khalaf ibn al-Mu’adh is found in an eleventh-century Latin manuscript; although the instrument bears no date it is clearly very early, not least because the plates serve the seven climates of Antiquity, as do the earliest Eastern Islamic astrolabe plates following Byzantine models. A single Andalusi astrolabe survives from the tenth century and is preserved in the British Museum (inv. no. OA+371), London; it is neither signed nor dated. Both are modelled after the Eastern Islamic astrolabes of the eighth-tenth centuries (all of which are published in King, In Synchrony with the Heavens, II, pp.403-544).

Otherwise there are some fifteen surviving Andalusi astrolabes from the eleventh century but none at all from the twelfth century. (For a list of all known Islamic astrolabes from before c.1500 see King, In Synchrony with the Heavens, II, pp.993-1020).

All numbers on the instrument are expressed in the standard Arabic alphanumerical system known as abjad (on this see Irani, “Arabic numeral forms’’). They are written in a distinctive unpointed sheriffed Western Arabic Kufic script, and even persons familiar with Arabic will falter at the sight of the year-number tâ’-ya’-alif = 400+10+1 = 411 in the dedication.

What is an astrolabe?

Imagine the heavens as fixed on a sphere centred on the observer. The heavens appear to rotate about a celestial axis once in a period that we divide into 12 hours (or 360°) and thus individual celestial bodies appear to move parallel to the celestial equator. The astrolabe is a two-dimensional representation of the three-dimensional celestial sphere in the plane of the celestial equator. This is achieved by a mathematical projection known as stereographic. The rete or rotatable pierced frame bears pointers for significant stars and an eccentric graded circle for the ecliptic or apparent path of the sun against the background of stars. The rate can rotate over any of a series of plates for a series of preferred latitudes. One observes a star or the sun and sets the appropriate pointer or solar position on the ecliptic scale on top of the appropriate altitude circle on the plate below, and obtains immediately the instantaneous configuration of the heavens with respect to the local horizon and meridian. With this one is armed for over one thousand possible operations that the astronomer al-Sûfî outlined in his tome on the instrument. It is hardly surprising that the astrolabe was the favourite instrument of Muslim astronomers, although it was not the only one.

The interested reader should consult www.astrolabes.org or King, In Synchrony with the Heavens, II. There is much false information circulating about the astrolabe, namely, that it was used in navigation, that it was used with the moon or planets, that one can compute eclipses with it, that one can use it to find the qibla or local direction of Makkah, that one can determine horoscopes with it, and more.

All numbers on the instrument are expressed in the standard Arabic alphanumerical system known as abjad (on this see Irani, “Arabic numeral forms’’). They are written in a distinctive unpointed sheriffed Western Arabic Kufic script, and even persons familiar with Arabic will falter at the sight of the year-number tâ’-yâ’-alif = 400+10+1 = 411 in the dedication.

The throne

The unusual throne resembles a seal sitting up on its haunches, with its face at the round pin holding the shackle and ring (both original). Two diagonal lines intersect at the vertical line.

The mater

The outermost scale of the mater is divided and labelled for each 5°, with subdivisions for each 1°, clockwise up to 360°. The astrolabic markings inside the mater serve latitude 66°, where the longest day is 24 hours (see below on the plates).

The plates

The six plates bear altitude circles marked and labelled for each 6° and azimuth circles above the horizon for each 10°. On each set of markings below the horizon from right to left we find in addition to the curves for the seasonal hours, specials curves that would serve for regulating the times of Muslim prayer. The inscriptions identify the curves and their functions: al-maghrib (sunset), maghib al-shafaq (nightfall), al-zawal (meridian), awwal al-zuhr (beginning of the midday prayer), awwal al-‘asr (beginning of the afternoon prayer), akhar al-‘asr (end of the afternoon prayer), tulu’ al-fajr (daybreak), and al-mashriq (daybreak). Additional inscriptions on the altitude circle for 18° – al-shafaq on the right and al-fajr on the right – indicate that this curve can also be used to determine twilight (in conjunction with the point of the ecliptic opposite the solar longitude).

For each side of each plate the inscriptions identify one or two localities corresponding to the latitude for which the plate is intended. The hours and minutes of the longest day at each latitude are given, these having been calculated from the latitude using the obliquity of the ecliptic, in this case, the value 23;51,20°, derived by Ptolemy of Alexandria c.125 AD. Whilst Muslim astronomers in Baghdad in the early ninth century had derived the much better value 23;35°, Andalusi instrument makers and astronomers in general remained somewhat tradition-bound.

19°             Hejaz, Yemen               13;10h
21;40         Mecca                           13;24
25              Madina                          13;35
27;30         Hijaz Misr                      13;46
30               Cairo                            13;58                           
33               Qairawan                      14;  8
34               Damascus                    14;18
37               Malaga                         14;36
38;30          Cordoba                       14;45
40               Toledo                          14;54
42               Saragossa                     15;  8
44                        -                          15;23
66               [Arctic circle]               24

It is interesting to compare the two different selections on the Edinburgh and Berlin astrolabes, which range from Ghana to Samarqand. All geographical data on early Islamic and medieval European astrolabes is available for comparison (King, 'Geography of Astrolabes'). 

The back

The back features a circular depression of 10cm and 2.5mm depth concentric with the outer rim. It is by no means clear why this is the case. To start with, the space would not have served to fit a removable plate or two, of the kind we find in later Andalusi astrolabes of the eleventh century, namely, one fitted, say, with a universal plate (see ‘Shakkaziyya’ in Enc. Islam 2); we can assert this not least because such plates were not invented until several decades later. If there had been a plate or two then the alidade would not have fitted in the ensemble. As it is, the ends of the alidade are cut so that the extremities move around the outer scales and the body of the alidade rotates over the empty ‘mater’. Furthermore, there are no original markings on this back ‘mater’ except for a square frame and the inscription within a rectangle at the top thereof:

'Made by Muhammad ibn al-Saffar in the city of Cordoba in the year 411 (1020-21 AD).'

The outermost scale of the back is divided and labelled for each 5°, with subdivisions for each 1°, arranged as four altitude scales each running up to 90°. There are notches at each of the 5°-divisions. Within this is a second scale, in twelve parts, divided and labelled for each 5° up to 30°. The names of the corresponding zodiacal signs are engraved within the mater, counter-clockwise from the right-hand side, and immediately within these are the names of the appropriate Julian months, arranged around a very crude scale for the days of each month. The casual way in which these scales have been executed is very surprising, but there is no reason to suppose that they are not original.

There are incompetent markings for a double shadow square to base 12 below the horizon on the back mater'. The divisions are carelessly and incorrectly engraved. The numbering of the subdivisions on the scales (in alphanumerical notation) indicates that these were added still in medieval times. The problem with these markings is that they fit perfectly within the square around the inscription. In other words, they do not seem to be a later addition.

The rete is a replacement from the Islamic East, either Egypt or Syria, or more probably Turkey, dating from the sixteenth or seventeenth century. It is typical in design of rates from those regions: simple and practical and without adornment. The back of the rete shows signs of having been reworked and modified. The pointer for the star Regulus, which is close to the ecliptic, is at Leo 24°, which corresponds to an epoch of c.1550. The rete fits perfectly within the mater of Ibn al-Saffar’s astrolabe. Likewise, the replacement rete from Iran which is found on Ibn al-Saffar’s Edinburgh astrolabe fits perfectly. Both astrolabes, now fitted with replacement retes with updated star-positions, could have had a new lease of life with the appropriate plates.

There are pointers for 6+8+5+7 = 26 stars, named as follows (counter-clockwise from the vernal equinox on the left, and arranged in four quadrants):

/1/ batn al-qaytus / al-ghul / al-dabaran / al-‘ayyuq / rijl al-jawza / mankib al-jawza /2/ al-shi’ra al-yamaniya / zahr al-asad / rijl (al-dubb) / ra’s al-asad / ‘unuq al-shuja’ / qalb al-asad / zahr al-dubb / janah al-ghurab /3/ al-simak al-a’zal / al-ramih / al-fakka / dhanab al-‘aqrab / al-jathi /4/ (al-)nasr al-waqi’ / (al-)nasr al-ta’ir / dhanab al-dajaja / mankib al-faras / dhanab al-jady / unlabelled on horizontal bar  / dhanab qaytus

We are indebted to Professor David King for providing the above catalogue note.


Enc. Islam 2, articles 'Ibn al-Saffar', 'Asturlab', 'Mizwala', 'Zidj'
Enc. Islam 3, article 'Astrolabes, quadrants and sundials’
B. R. Goldstein, article 'Ibn al-Saffar' in Enc. Islam, 2nd edn.
A. K. Irani, 'Arabic numeral forms', Centaurus 4 (1955): 1-12, repr. in Kennedy et al., Studies, pp.710-721. S. Kennedy, colleagues and former students, Studies in the Islamic Exact Sciences, eds. M. H. Kennedy and D. A. King, Beirut, 1983.
Sa’id al-Andalusi, Tabaqat al-umam, ed. L. Cheikho, Beirut, 1913, trans. R. Blachère, Catégories des nations, Paris, 1935.
D. A. King, 'Three sundials from Islamic Andalusia' (1978), repr. in idem, Islamic Astronomical Instruments, London, Variorum, XV.
D. A. King, 'The geographical data on early Islamic astronomical astrolabes' (1999), repr. in idem, In Synchrony with the Heavens, vol.2, pp.915-962.
David A. King, In Synchrony with the Heavens – Studies in Astronomical Timekeeping and Instrumentation in Medieval Islamic Civilization, 2 vols., I: The Call of the Muezzin, II: Instruments of Mass Calculation, Leiden & Boston, Brill, 2004-05.
L. A. Mayer, Islamic Astrolabists and their Works, Geneva, Albert Kundig, 1956.
R. W. Plenderleith, 'Discovery of an old astrolabe', Scottish Geographical Magazine 76 (1960): 25.
F. Sezgin, ed., Arabische Astronomische Instrumente, 6 vols., Frankfurt, IGAIW, 1990-91.
M. Rius, article 'Ibn al-Saffar, Ahmad …', in Th. Hockey et al., eds, The Biographical Encyclopedia of Astronomers, New York, Springer, 2007, pp.566-7.
H. Y. Sheynin, 'Claudius Ptolemy? Pseudo-Ptolemy? The Main Source of Moses Almosnino’s Treatise on Astrolabe', Journal for the History of Astronomy 46 (2015), pp.343-350.
F. Woepcke, 'Über ein in der Königlichen Bibliothek zu Berlin befindliches arabisches Astrolabium', Abhandlungen der Königlichen Akademie der Wissenschaften, 1858, pp.1-31, repr. in Sezgin, ed., Arabische Astronomische Instrumente, vol.2, pp.1-36.