1580s Iberian astrolabe (reproduction)

The 1580s Iberian astrolabe reproduction.
Figure 1: The 1580s Iberian astrolabe reproduction.
In 2006 I met clockmaker GŁnther Oestmann of Ars Mechanica at the National Maritime Museum in Greenwich. He then had a talk on casting an astrolabe, which he had recently done. After the talk we discussed the possibility to create an Iberian astrolabe for my collection. The museum kindly supplied details of the Valentia astrolabe (NAV0022) in their collection, which I used to create an AutoCAD drawing (see figure 14). This I subsequently sent to GŁnther who then had the body cast and built the instrument. During the A Sense Of Direction conference in May 2010 we met again and I was presented the astrolabe.

The mariner's astrolabe was one of the first actual altitude measuring devices at sea. Its predecessors were the quadrant and the kamal. Earlier quadrants were however not yet used with a limb divided in degrees, but - alike the kamal - were used in a relative way with marks for the individual ports. The mariner's astrolabe developed during the 15th century out of the astronomical astrolabe and its earliest recorded use at sea dates from 1481 when it was used in a voyage by Diogo d'Azambuja down the west coast of Africa.1

Eventually the instrument was replaced by the cross-staff and backstaff instruments like the demi-cross, the hoekboog (double triangle) and finally the Davis quadrant.

Mariner's astrolabes were generally made of bronze or brass, but wooden ones are known to have existed.2 The one created for me was cast from a modern bronze alloy (CuSn10-C) to distinguish it from an original, is 182 millimetres in diameter, has a bronze alidade with brass vanes and a brass hinge (see figure 5) and weighs approximately 2.7 kilograms, slightly more than the original. The alidade is held in place by a pin (see figure 6) that is locked by a key or horse (see figure 4) similar to 16th century examples (in the 17th century a wing-nut was more commonly used).3 The vanes are pierced for sun observations only (see figure 3). The limb is marked for zenith distance (see figure 5).

The 1580s Iberian astrolabe reproduction from the other side.
Figure 2: The 1580s Iberian astrolabe reproduction from the other side.
As observations are done relative to the local vertical (see figure 7), there is no need for a horizon to function. This allows the instrument to be used at any place as long as the sun or polar star can be seen from it. In practise this meant that - for sun observations - the instrument could be used below deck near the vessel's centre of gravity with the sun visible through a hatch. In this way the astrolabe (and the observer) was less affected by lateral movement, thus making observing easier and more accurate. The observer would sit down, rest his arms on his knees, suspend the instrument between his legs and take his observations.

The first observations with the reproduction were done outside the National Maritime Museum at the colonnades to the Queens House on May 7th 2010 around noon. From these observations a 20 arc minute instrumental error was found (between face left and face right), which was later corrected. The average was no more than 3 arc minutes off (3 nautical miles).

On May 14th 2010 another test, consisting of 100 observations in my sheltered back garden, showed that the instrumental error was adequately dealt with. Around 3 arc seconds, the instrumental error between face left and face right is now negligible, although an error between the pointers remains of about 3.6 arc minutes. The latter - just as well as the instrumental error between the faces - will however average out when observing in two faces and with both pointers as one would do when observing with a 1962 Wild T2 or 1919 K&E.

The only errors left are scale and observer's errors (see figure 11), in this case still around 10 arc minutes (0.2mm average displacement of the projected sun on the lower vane). The standard deviation calculated over all observations was 10 arc minutes as well (1σ, 68%).

The pinnules in the vanes of the alidade.
Figure 3: The pinnules in the vanes of the alidade.
In order to investigate the cause of this average error I took another series of measurements. This time I observed in two different fashions: forward and backward, e.i. I started with facing the sun and looking at the astrolabe from above, taking eight observations.

Then I took eight observations with my back towards the sun and holding the astrolabe at about eye-level so I could look at the alidade from an angle mirrored to the one in the forward fashion (see figure 9). What I suspected came out: when observed backward I get readings too low, when observed forward they are too high (see figure 12).

The cause of this deviation lies in the shape of the pinnule. Due to the drilling process a small burr will appear around the pinnule. In order to get the hole smooth again the burr was taken off using a countersink. This resulted in a small bevel along the edge of the hole (see figure 10). Although minute and not visible to the naked eye at only 0.15mm, this is large enough to cause a 7 to 8 arc minute observer's error.

The key that lockes the pin of the alidade.
Figure 4: The key that lockes the pin of the alidade.
Whether or not period instrumentmakers used the same technique to get rid of the burr is (yet) unknown, but if they did it in the same way (the alternative is to sand it off) then period instruments would have the same problem. As the 16th and 17th century navigators would only observe in a forward fashion, this would then cause them to get values too high for the sun's altitude. Which on itself would result in latitudes too low and hence a course steered too high when running down the latitude. From this second test an average error of 4 arc minutes with a standard deviation of 13 arc minutes (1σ, 68%) was calculated. The instrument error was again found to be less than one arc minute, with 3.4 arc minutes between the pointers.

After much deliberation I decided to remove the bevelled edge by drilling the hole to 1.3 millimetres (it was 1.0 millimetre originally). It immediately became clear that the instrument performed much better. A series of 40 forward observations resulted in an average error of 4 arc minutes with a standard deviation of 8 arc minutes (1σ, 68%). This proves that John Davis was right that he stated that '...it is an excellent Instrument, being rightly understood and operated...', at least on land that is.4

I'm very grateful to Richard Dunn National Maritime Museum in Greenwich) as without his details of the Valentia astrolabe in their collection the reproduction of this instrument would not have been possible. Of course I want to thank GŁnther Oestmann of Ars Mechanica for creating the astrolabe.

Notes

[1]: A. Stimson, The Mariner's Astrolabe, A Survey of Known, Surviving Sea Astrolabes, (Utrecht, 1988), p. 16.
[2]: idem, pp. 15-16.
[3]: idem, p. 24.
[4]: J. Davis, The Seamans Secrets, Devided into two parts, wherein is taught the three kindes of Sayling, Horizontal, Paradoxal, and Sayling upon a Great Circle. Also an Horizontle Tyde-Table for the easie finding of the Ebbing and Flowing of the Tydes, with a Regiment newly Calculated for the finding of the Declination of the Sun, and many other most necessary Rules and Instruments not hereforte set by any. Newly Corrected and ammended, and the Eight time printed., (London, 1657).


If you have any questions and/or remarks please let me know.

The hinge of the astrolabe.
Figure 5: The hinge of the astrolabe.
 
The notch below the pin has an extra cut to distinguish the pointers.
Figure 6: The notch below the pin has an extra cut to distinguish the pointers.

Holding the astrolabe for observations.
Figure 7: Holding the astrolabe for observations.
 
A spot of light from the upper pinnule has to be cast on the lower.
Figure 8: A spot of light from the upper pinnule has to be cast on the lower.

The difference between the forward and backward observation methods.
Figure 9: The difference between the forward and backward observation methods.
 
A bevelled edge to the one millimeter diameter pinnule resulting in a 0.15mm or 7-8 minute error.
Figure 10: A bevelled edge to the one millimeter diameter pinnule resulting in a 0.15mm or 7-8 minute error.

More than an hour of observing in Face Left , Face Right, pointer zero and pointer one.
Figure 11: More than an hour of observing in Face Left , Face Right, pointer zero and pointer one.
 
A series of 8 backward and forward observations showing the influence of the bevelled pinnules.
Figure 12: A series of 8 backward and forward observations showing the influence of the bevelled pinnules.

After removing the bevelled edge from the pinnules the instrument performs much better.
Figure 13: After removing the bevelled edge from the pinnules the instrument performs much better.
 
The drawing of instrument based on the original in the NMM.
Figure 14: The drawing of instrument based on the original in the NMM.

Celestial Navigation Coastal Navigation Distance measurement
1580s Mariner's astrolabe 1590 Hood's cross-staff 1618 Demi-cross 1623 hoekboog 1660 spiegelboog 1661 Kronan cross-staff 1720 Hasebroek cross-staff 1734 Davis quadrant Early 19th c. ebony octant Late 19th c brass octant 1941 U.S. Navy quintant Hirado navigation set