Recently in Manuscript Category

Installing DigiPal - Part 2, Scripts and Configuration

Once you have all the parts installed, there is some configuration that the Dockerfile takes care of automatically that you will have to do by hand


(The AWS-way to do this would be to spin up a Postgres RDS instance, but that was too many variables for a first try. I leave it as an exercise to the reader)

  • Login to postgres and create a digipal DB and digipal user, for that the instructions in GitHub are accurate, see the databasesection, until the step that starts "After that, run in your terminal the following commands:". 

  • By default postgres is not configured to allow password login locally, only "account" based socket login. You need to change that by editing /var/lib/pgsql93/data/pg_hba.conf. Set the line that looks like 

local   all             all                                   ident

to be

local   all             all                                     md5

Replace ident with md5 and restart postgres. You can now continue in the GitHub instructions from "After that, run in your terminal the following commands:"


  • Add to /etc/lighttpd/lighttpd.conf, at the endinclude "vhosts.d/digipal.conf"
  • uncomment in /etc/lighttpd/modules.confinclude "conf.d/fastcgi.conf"
  • create /etc/lighttpd/vhosts.d/digipal.conf
#include_shell "/usr/share/lighttpd/"
#include_shell "/usr/share/lighttpd/"
fastcgi.server = ( "/iip/iipsrv.fcgi" =>
  (( "host" => "",
     "port" => 9000,
     "check-local" => "disable",
     "min-procs" => 1,
     "max-procs" => 1,
     "bin-path" => "/etc/lighttpd/iipsrv.fcgi",
     "bin-environment" => (
        "LOGFILE" => "/tmp/iipsrv.log",
        "VERBOSITY" => "10",
        "MAX_IMAGE_CACHE_SIZE" => "20",
#        "FILENAME_PATTERN" => "_pyr_",
        "JPEG_QUALITY" => "75",
        "MAX_CVT" => "3000",
        "FILESYSTEM_PREFIX" => "/apps/digipal/images/"

NB. you will need to set the FILESYSTEM_PREFIX as needed for your environment

  • Copy iipserv.cgi to /etc/lighttpd/ and set it as executable: chmod a+rx /etc/lighttpd/iipserv.cgi


As I have my system configured these apps are not accessable to the outside world, but only to localhost. When I want to use it, I ssh to the machine and do local port forwarding, i.e.ssh -L 8000:localhost:8000 -L 8081:localhost:8081 ${USER}@${SERVER}The port 8000 is used for the web server, 8081 for the image server (that's what lighttpd does). You should then be able to access your digipal by pointing a browser to http://localhost:8000

NB. this worked for me, after some hacking around, but I provide no guarantee that these instructions are complete.  Moniti estis

Digipal in AWS, Part 1

| No Comments | No TrackBacks

Installing DigiPal in AWS

There is some documentation on spinning up Digipal in a server, versus the Docker distribution, on their Github, but I found that information incomplete. Here is what I had to do to configure a running instance in AWS, running on AWS's linux AMI, which is mostly like Fedora. Note that I do not have Nginx running, so there's still some port weirdness. This is part 1 of 2, we'll just install and build everything. Part 2 will have configuring lighttpd to serve the images and postgresql account stuff


What to install

The following packages all to be installed with yum. There's a single-line at the bottom if you want to just do it

  • git - git (duh)
  • Postgres - postgresql93 postgresql93-server postgresql93-devel
  • ImageMagick - (note that we only need the CLI tools, not the devel libs) ImageMagick
  • libmysqlclient-devel - (stated in the reqs, no idea why) mysql56-devel
  • lighttpd - (needed for iip fastcgi) lighttpd lighttpd-fastcgi
  • Pre-reqs to build IIP - gcc, gcc-g++ libxml2-devel libxml2-python27 libxslt-devel automake autoconf libtool libjpeg-devel libtiff-devel
  • Lessc - (used by Digipal in the CLI, runs in Node.js. Note that this is not part of the AWS repos, so you have to pull from epel) nodejs npm python-lesscpy

Big omnibus commands, need to be run as root or via sudo

yum install git postgresql93 postgresql93-server postgresql93-devel ImageMagick 
mysql56-devel gcc, gcc-g++ libxml2-devel libxml2-python27 libxslt-devel automake
autoconf libtool libjpeg-devel libtiff-devel lighttpd lighttpd-fastcgi
yum install nodejs npm python-lesscpy --enablerepo=epel

What to build

  • lesscp - needs to be built by npm, -g makes it global - npm install -g less


  • Get the code from git - git clone
  • Build it:
cd iipserv
make check
sudo cp src/iipsrv.fcgi /etc/lighttpd/


Now time to install digipal itself

  • get the code from Git
git clone
cd digipal
git checkout 1.2.1a
  • Now follow their instructions to get dependencies and install
    pip install -r requirements.txt

NB 1.2.1a is the current version in git. If you want to look at the versions, go to the GitHub page and click the "branches" dropdown. Replace 1.2.1a with any of those values verbatum

This horribly mangled item is the remains of what once was probably a beautiful Book of Hours, probably from Spain.  None of the miniatures or historiated initials remain, but what does survive includes the February-December of the Calendar, a piece of the Litany and terce-vespers (incomplete on both ends and the start of each hour) of the Hours of the Virgin.  The scribe who wrote this was extremely careful and talented.  Every page is ruled in red ink, double lines to the edge of the membrane for the edge of the text-block and a single line within the text block for each line of text

Cambridge-Harvard Houghton Library-Typ 731 - 12r.jpg


For the calendar pages, the ruling is more complex, with two additional double-lines, though not extending above or below the text block, setting up columns for the Dominical Letters and the Golden Numerals.

Cambridge-Harvard Houghton Library-Typ 731 - 5r.jpg


Besides the care in ruling and preparing, the scribe's writing is exceptional.  The strokes are even in both color and width and the text block is filled cleanly and evenly with no odd hanging strokes or unfilled lines.  Even the letter spacing is even

Cambridge-Harvard Houghton Library-Typ 731 -_.jpg

(Terce of the BVM from Typ 731)

Though none of the miniatures survive, the profusion of colored and gilded initials inline, particularly the Litany, suggest that this would have been a lavish book when intact

Cambridge-Harvard Houghton Library-Typ 731 - Litany.jpg

(Litany from Typ 731)


Black before Red

| No Comments | No TrackBacks

Sometimes errors provide an interesting insight into processes. In this case the context is the calendar section of a Book of Hours held by the Boston Public Library, MS q. med 89, which dates to the mid 15th century and was made in Poitiers, France. In this calendar there are three columns of text:

  • The main feast or saints name (black text for most, red for the most important)
  • A dominical letter, used for calculating the date of Easter (black text)
  • The numeral of the Roman Date, i.e. VI. (Red text)

There are 2 interesting errors that point to the practice of writing all of the black text first and then going back to add in the red. On the verso of May the Roman number is offset one row versus the Feast and the Dominical, which are in alignment (f.6v below) Boston-Boston Public Library-MS q med 89 - 6v02838.jpg Notice that the red roman numerals start one row down from the top and continue one line past the black text. The dates of the feasts match the dominical letters and are offset from the Romans exactly by one.

The second one is at the very end of the verso of of November (f.12v below). In black we have, on November 29th, ‘Vig’ as shorthand for “The Vigil of St. Andrew, apostle”, and on the following page, 7 days later on the 7th of December, the Octave of the same Andrew(see f.13r below). There is, however nothing on November 30th, no feast at all. This should be the main feast of St. Andrew, and as the feast-day of an Apostle, it should be in red. The presence of both the Vigil and Octave, on the proper dates, is strong evidence that the saint should have been there and was forgotten, rather than being intentionally omitted. I would propose that the scribe simply forgot the last line on the page when he was doing the second pass for the rubrics. Boston-Boston Public Library-MS q med 89 - 12v02850.jpg Boston-Boston Public Library-MS q med 89 - 13r02851.jpg


Typ 295, from the Houghton Library at Harvard University, is a late 13th Century single-volume bible.  Specs are: 368 leaves of vellum, very thin, about 18cm high. It contains the Vulgate text with some additional apparatus, in this case Interpretationes Hebraicorum nominum and a table of readings.  These are often referred to as Parisian bibles, due to the large number of them which were created in the Paris stationary trade, but the catalogue identifies this on as Italian, due to the roundness of the gothic hand. There's plenty to say else about the manuscript, but I'm focusing on the rulings on the page.

Genesis 36.jpg

The rulings consist of three parts: Titles, Text and Decorations.  Moving from top to bottom, the first pair of lines locate the title of the page, in this case "Sys" (the facing page has "Gen").  The next pair of horizontal lines mark out the first line of text in the block, and the rest of the block has ruling only within it, not extending to the edges of the page.  Finally there's a horizontal pair near the bottom of the page, which does not appear on all the pages of this MS.  I'm guessing that it has to do with the decorations, but that's speculation. 

For the vertical rulings, moving from the inside out, there's a pair of vertical lines to define the inside margin of the text block which also serve to structure the red/blue penwork decoration.  Next there's a double-column ruling, three lines, to spit the columns of text.  There's a vertical pair to end the text block and one final one to mark out the title margin, in this case for XXXVI.  This final block also contains penwork on some pages. 

All of these rulings, three horizontal pairs and four vertical sets, extend nearly to the edges of the membrane, far beyond where they are used.  In contrast the text block itself is ruled internally only, crossing the center gutter but not extending outside the margins after the first line.  The text floats between the rules lines, neither sitting on the baseline nor hanging from above.  Some descenders cross the baseline, but only a small amount (see closeup of Genesis 35, below).  The scribe has fit an enormous amount of text on the page, from Genesis 34:19 through part of 35:23, or about 1000 words

Genesis 35 - closeup.jpg

The last section of the manuscript, the Interpretationes Hebraicorum nominum, attributed to Jerome, but probably written later.  For our purposes it is interesting because the scribe has shifted to three columns of text.  These are demarcated only by 4 vertical rulings, the one to the left of each column also serving to structure the constant list of capital letters (see 'B' below)

Hebrew Names B.jpg

All of the rulings appear to be in ink

De Bellis Macedonicis is one of those manuscript fragments more important for what it heralds than what it is.  As a text it's just a bit of a history on a war between Rome and Macedonia, possibly written by one Lucius Arruntius, found with a vast number of papyrus fragments at Oxyrhynchus in the 19th century.  It is, however, currently the oldest known representitive of the most important western manuscript form, the parchment codex. 

Despite it being referenced and reproduced quite often in Paleography textbooks, the British Library had not, until this past week, publically digitized it.  Despite the clarity of the letters, the text isn't all that easy read.  The letters are a mix of Rustic (A, M, U), Old Roman Cursive (q and p) and Uncial (D,F,R).  The Old Roman Cursive is particularly strange to modern eyes, see the word "Imperi..." at the end of the first line. 

The fragment was dated to the 3rd Century by Lowe, but has more recently been dated to the end of the 1st C CE by Michelle Brown (A Guide to Western Historical Scripts from Antiquity to 1600. London: British Museum, 1990; repr. 1993. [52 plates])


American Qu'ran

| No Comments | No TrackBacks

I'm a bit late to the party, but a sample of the American Qu'ran created by Sandow Birk.  It's a complete manuscript of the translated text with illustrations of contemporary American life.




Where Babies Come From...

| No Comments | No TrackBacks

The British Library has finally digitized MS Harley 4425, their profusely illuminated 15th Century copy of Romain de la Rose.  Within it is the answer to the age-old question "where do babies come from?"  The answer, of course, is "babies are forged by nature in a smithy"



Ottoman Miniature Movies

| No Comments | No TrackBacks

Redoing movie posters seems to be a thing on the internet, however I've rarely seen them done in a style that required such skill and attention to detail as these in the style of Ottoman miniatures by Murat Palta.

The GodfatherScarface
Star WarsThe Shining
star wars.jpgthe shining.jpg
Kill Bill


Comet Sketches

| No Comments | No TrackBacks

Argentum im Zeichen des Jupiters


The 'Kometenbuch' (Comet Book) was produced in 1587 in the area of Flanders or North-East France. This is one of the many illustrations (see the source for more), in this case a comet of silver hue passing through the region of Jupiter.  A digitized version is available from the Universitätsbibliothek Kassel


About this Archive

This page is an archive of recent entries in the Manuscript category.

AWS is the previous category.

Find recent content on the main index or look in the archives to find all content.

OpenID accepted here Learn more about OpenID