[ISHMap-List] A critical review of the hypothesis of a medieval origin for portolan charts

Roel Nicolai rnicolai at xs4all.nl
Mon Mar 17 17:36:14 CET 2014

Dear all,


I have just registered as a member of this forum and cannot yet post
directly, therefore an email to provide some initial answers.


a)      Yes, I was awared the PhD degree by the university.

b)      Regarding the availability of my thesis: I am trying to get a trade
edition published, so for that reason the pdf file is not (yet) available.

c)       Luis A. Robles Macias gives a good summaryof the press release, but
one point is not correctly reflected: "they were much further advanced in
terms of knowledge in the Middle Ages than we think". This is part of the
following longer sentence: "Nicolai concludes that part of history needs to
be rewritten. "This needs to happen even if I am wrong, because that would
mean that they were much further advanced in terms of knowledge in the
Middle Ages than we think." 

d)      Matthew Edney's response is very much to-the-point. PR people indeed
go for the spectacular and may even have a dislike for cautious wording! The
numercial analysis of five charts (and the Compasso de Navegare) is only
part of the thesis and indeed should never stand on its own. The historical
context is absolutely indispensible. However I would like to reverse this
statement: without taking into account the extraordinary geometric aspects
of the charts a (purely) historical explanation of their origin and
construction will not be adequate.  

e)      Joaquim Alves Gaspar states that the Mercator projection is old
news, which it is. In 1987 Scott Loomer found that the Mercator projection
is the projection that fits best to the charts but he analysed them as
single charts. I analysed them as composite charts and found a mix of best
fits to the Mercator projection and the Equidistant Cylindrical. This last
type of projection features in the press release as "or similar type of
projection". Nevertheless, the (sub-)charts share more characteristics with
the Mercator projection than with the Equidistant Cylindrical.

f)       I may have missed Ramon Pujades' change of heart since his book
"Les cartes portolanes" if he now favours the beginning of the 13th century.
But unless he has uncovered something fundamentally new, I doubt that this
has a serious impact the results of my study, but if it has, please let me
know! Having a demanding full-time job I have concentrated the last few
years on writing my thesis in the evening hours and weekends and as I am not
part of the regular network of map historians I may have missed a few recent

g)      I disagree with Gaspar's conclusion that the magnetic declination of
around 1200 is reflected in the charts and am certainly willing to discuss
my reasoning. By the way, I (also) used the CALS7k.2 model, but evaluated it
for 1250.

h)      The accuracy of the portolan charts is consistently high, but only
if they are treated as composites. The high accuracy holds for any
sub-chart, but the composite indeed reflects regional scale differences and
orientation differences.

i)        The question whether the map projection is a deliberate part of
their construction or just accidental (Gaspar's portal entry) is not easy to
answer, but is absolutely key to how we interpret those charts. The short
answer is that I constructed three networks of bearings and (rhumb line)
distances in the Western and Eastern Mediterranean and the Black Sea, based
on what were likely routes to sail given the prevailing summer winds in
these waters. I treated these 'synthetic' bearing and distance pairs as free
from observational errors and processed the three networks as if the earth
were flat, i.e. by plane charting. This resulted in a series of coordinates
of the points of each of the three networks, obtained by plane charting and
subjected each of the three plane charted network to the same cartometric
analysis as I used for the portolan charts. If these networks naturally
'contained' the Mercator projection (or the Equidistant Cylindrical), the
cartometric analysis process would yield a Mean Squared Error (MSE) very
close to zero, i.e. a value small enough to 'hide' in the accuracy
characteristics of the charts. This turns out NOT to be the case: the MSE
'due to plane charting'  is close to or larger than the accuracy figure of
the real charts. That is an algebraic impossibility. Ergo: the map
projection is not an accidental by-product of plane charting.


Best regards,

Roel Nicolai

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lazarus.elte.hu/pipermail/ishm/attachments/20140317/f23035ce/attachment.html>

More information about the ISHM mailing list