Page 1 of 1

Gutenberg and ligatures.

Posted: Thu Sep 01, 2005 6:44 pm
by William
Some time ago I wrote the first part of the following and posted it at the decode Unicode webspace.

That webspace is now at on the web.

The item is part of the English version and is under Private Use Area.

I have just added the supplementary note.

The Chronicle Text font has its own thread in the Gallery section of this webspace.



You might find the following of interest. It is the web information of a television program in a series called Renaissance secrets which was on the BBC in England some time ago.

---- start of additional note ----

Additional note of Friday 18 April 2008.

The above link does not seem to work anymore. However, the following will provide the information at the time of writing this additional note. ... opsis.html

---- end of additional note ----

An interesting point is as to why Gutenberg had so many ligatures! With metal type in the twentieth century the making of an additional ligature was a lot of extra work in that an additional metal punch to make metal matrices had to be produced. Today, with electronic fonts, the constructing of an extra ligature glyph also takes extra work.

Yet was the same true for Gutenberg? Did ligatures in fact save Gutenberg work?

My reason for thinking that this is a possibility is as follows. In that television program about Gutenberg a researcher had made images of individual characters printed by Gutenberg and found that each character of a sort had small differences which meant that they could not have been made from the same matrix. It is possible that Gutenberg used matrices of whatever material such that the matrix was destroyed during the casting of a character, thus meaning that as many matrices had to be made as there were pieces of type: the invention of a reusable matrix being a later invention, perhaps by another printer.

So making a ligature character would mean producing at least one less matrix than would have otherwise been the situation! A two-letter ligature would save the production of one matrix, a three-letter ligature would save the production of two matrices. Gutenberg used many ligatures. Perhaps it was a money-saving idea as well as an artistic idea: thus, as one might say, "painting two birds on one canvas" in that one idea served two purposes!


Supplementary note of 1 September 2005

Earlier today I was adding some ligature glyphs into the Private Use Area of one of my fonts, a font named Chronicle Text. It is a black letter font.

I was adding a glyph at U+E70D within the font. It was not a ligature as such, but was two lowercase letters l, side by side. I was thinking about the above idea that Gutenberg may have used many ligatures to save work and it occurred to me that in that case maybe he cast often-used-pairs of letters onto one piece of type so as to minimize the number of pieces of type whilst increasing the number of sorts of pieces of type yet within the limits of only casting pairs, or even triples or longer sets, of characters for frequently used sequences which could be used in various pieces of text.

I then wondered whether Gutenberg thus might have used far more 'ligatures' than people have thus far detected, simply because those undiscovered ligatures are for items where there is no visible join between the two parts of the ligature.

Is that a hypothesis which could be tested by measuring the relative positions on the printed page of two characters which might have been so used, for each use of the pair of characters, bearing in mind that there might be several notionally identical sorts for the pair of characters?

William Overington

1 September 2005

Posted: Fri Sep 02, 2005 6:25 pm
by William
I have been thinking further on this and wondering which pairs of characters could reasonably form a ligature which has no visible join, that is, two characters cast on the same piece of metal yet which when printed print the same as if two separate pieces of type were being used.

I am thinking that perhaps combinations with a lowercase i would be a good possibility as some of these would be used fairly often and it would save having to cast so many thin characters. Would casting a thin i be harder than casting a wide m, due to metal flow problems?

I then thought that perhaps combinations with a lowercase l would be a good possibility too.

For both ligatures with lowercase i and lowercase l, such ligatures would perhaps not have a visible join between the two characters, except perhaps for fi and fl, both of which survive to the present time. So maybe items such as pi and ni were used. Suppose that one tries to design a pi ligature or an hi ligature, how exactly would it look? With a ligature such as ct or ch or ha or he or ho there is enough information in the shapes that the two characters can be presented with no gap between them. However, with pi or hi or ni or ol or similar, a lack of a gap could cause a legibility problem.

I am looking at the web about Gutenberg's printing.

Readers might like the following link.


Posted: Sat Sep 03, 2005 6:12 am
by Dick Pape
Hello William.

I thought ligatures were invented or used mostly for spacing considerations -- that is, a lig is narrower than two separate letters so would be used by the scribes of the day to fill out a line to the proper or full width.

It was thus carried forward to type printing to simulate handwriting as I understand also there was an initial prejudice against machine written texts.

There would be little need otherwise to have a specialized character(s) you would have to design, cast and inventory, you would think.


Posted: Sat Sep 03, 2005 9:28 am
by William
Dick Pape wrote:Hello William.

There would be little need otherwise to have a specialized character(s) you would have to design, cast and inventory, you would think.

Thank you for posting.

Indeed, I would agree, if the conventional model of the matrix being reusable is used.

However, the television programme of which the following webspace is the support pages, suggests that Gutenberg was not using reusable matrices, but matrices which were damaged as part of the process of producing a single piece of type. ... _info.html

The television programme was fascinating.

It seems to me that if the matrix is destroyed during casting, then having ligatures saves work, simply because fewer matrices then need to be made in order to produce a particular piece of text. It then becomes a balance between having more ligatures to save work in making the type balanced with only having ligatures that are likely to be used in many situations.

In order to try to get a simulation feel for the possibility using my Chronicle Text font in conjunction with WordPad I am hoping to add some hypothetical ligatures into Chronicle Text.

In order to get started, I am thinking that some p ligatures might be nice, so I am hoping to implement the p ligatures of and then add a pi ligature at U+E719 and a pl ligature at U+E71A.

Those ligatures will hopefully provide an opportunity for the hypothesis to be simulated.

Whether such ligatures were ever used is hypothetical, maybe the matter can be decided one way or the other using the techniques of the television programme.

Meanwhile, Font Creator 5 and the availability of the Private Use Area of Unicode makes a simulation possible, though only a partial simulation as each use of the glyph on the screen is from the same glyph of the font, so the differences due to different pieces of type of the same sort in the television programme cannot be simulated using this method.

However, it might be possible to devise a different simulation where differences due to different pieces of type are detectable. I am thinking that this would need a special, for-the-purpose font, with a type tray and its contents simulated within the Private Use Area using several almost identical yet slightly different glyphs for each sort.


Posted: Sat Sep 03, 2005 1:35 pm
by William
Chronicle Text version 0.23 is now available on the web.

The following additional ligatures are included in this latest version.

From the web page the following.

U+E760 ba
U+E761 be
U+E762 bo
U+E763 da
U+E764 de
U+E765 do
U+E766 ha
U+E767 he
U+E768 ho
U+E769 pa
U+E76A pe
U+E76B po
U+E76C pp
U+E76D va
U+E76E ve
U+E76F vu

U+E77A cha
U+E77B che
U+E77C cho

U+E77E ppe

I have also added hypothetical ligatures into U+E716 to U+E71B. These are bi, di, hi, pi, pl, vi respectively. These were decided upon so as to go conveniently with the ligatures above which begin with b, d, h, p and v.

People producing a simulation using WordPad using the Alt method for entering characters will hopefully find the following decimal values helpful.

E710 hexadecimal is 59152 decimal.

E760 hexadecimal is 59232 decimal.

E770 hexadecimal is 59248 decimal.

Whether the hypothetical ligatures were used in practice is unknown, yet hopefully the ligatures in the font will be helpful for trying a simulation and getting a feel for how it might have worked.

The Chronicle Text font has its own thread within these forums.


The Chronicle Text font is available directly from the following web page.


Posted: Thu Mar 09, 2006 9:02 pm
by sevry7
William wrote:
Dick Pape wrote:Hello William.

There would be little need otherwise to have a specialized character(s) you would have to design, cast and inventory, you would think.

Thank you for posting.

Indeed, I would agree, if the conventional model of the matrix being reusable is used.
You still see a lot of these ligatures in the mid to late 16th century printing. But by the early to mid 17th, many seem to have been removed from use, save for those which still seem preferable. And those would be the 'f i', 'f f' and so on. And the reason for that is simply that the dot on the, i, otherwise seems a little distracting. In the other, the 'f' seems to get crowded or seems incomplete. It might need to be padded or spaced out. With fixed type, you could slide vowels under a cap 'V' or 'W', say, and so on. But it would be a separate slug. And printers had hundreds and hundreds of these, at the ready, even if they didn't all quite exactly match. With the eventual adoption of the extended features of opentype, perhaps toward the end of the year (although Avalon is supposed to be XP compatible now), all these old tricks will come into vogue, again. The great advances made by the daisy-wheel printer and dot-matrix will only be an odd recollection. We'll have the advanced technology of the 16th century once again under our control and visible right there on the browser.

Posted: Fri Apr 18, 2008 1:50 pm
by William
Here is a link to a later thread named "Stephen Fry and the Gutenberg Press".


William Overington

18 April 2008