|February 15th, 2015
|history, music, tech
People have been distributing music in written form for at least 3000 years, but just as with text, the ability to make large numbers of copies of a written work in a relatively short amount of time has had large effects on the culture of music. The technology behind this massification is fascinating, with several methods of reproduction competing over almost 500 years.
Music, like speech, is linear and so generally amenable to writing. Also as with speech, there are many possible writing systems. The earliest known system for writing music is analogous to a syllabic writing system. Instead of indicating a combination of consonants and vowels, however, a symbol indicates a series of notes that are customarily played together  We see this writing on tablets from the 13th century BCE, discovered in the dig at Ugarit, in Syria.
We also have written music from Ancient Greece. Their writing system is more analogous to a phonetic writing system, with symbols for individual notes. These characters, like our letters, were completely symbolic. We know that most musicians in Greece did not use this system, however, and instead learned tunes by ear and with person to person instruction. The writing was mostly for philosophical/mathematical texts.
The Greek system of musical writing was adopted by the Romans as well, where it was used again mostly for philosophical writings. The main Roman text on music we have is Boetheus' de institutione Musica, a combination translation, harmonization, and expansion on earlier Greek sources. It is very much a theoretical work, and he writes "A musician is one who has gained knowledge of making music by weighing with the reason, not through the servitude of work, but through the sovereignty of speculation."  There were many people who played music and had learned by the 'servitude of work' but if any common musicians wrote things down none of it has survived.
Early medieval musical notation is the first system we still have which was used by musicians. In the 8th century monks began to write neumes, shapes that represented pitch contours, above words to serve as reminders of the melody. Like the Ugaritic system this is a syllabic system, but unlike the older (and unrelated) system it is not abstract. The height of a mark on the page indicates its relative pitch. Also unlike (the main interpretation of) Ugaritic notation, neumes did not unambiguously indicate the melody.
Over time people expanded on the neumes by writing them varying distances above the words they went with. While neumes had previously indicated pitch relative to the previous note, now they began to indicate absolute pitch. They still did not encode everything about the melody, but they were closer and perhaps could serve as a guide for people learning it.
The idea of a staff grew out of putting the neumes at varying heights. Drawing a line or several to make distances more clear became more and more common. In the early 11th century Guido d'Arezzo wrote a very popular (among monks) text, the Micrologus, which among other things included staff notation that told the melody accurately. It is likely that the idea of having the space between a line indicate twice the distance between notes in the scale did not originate with him, but his book popularized and standardized the notation.
Part of why this new notation spread so well was that it could do two new things: encode melody for people who did not already know it and encode multiple simultaneous notes for harmony. The Micrologus, in fact, is a treatise on harmony, and may have served to popularize both the writing system and an application that made it necessary.
The church system was now capable of encoding melody fully, but remained limited in rhythmic capacity. Sometimes the spacing between notes would roughly indicate the time between them, like on a piano roll, but generally the writing didn't make any attempt to communicate rhythm. For the traditional monastic chants and songs this was not a problem; they usually had rhythms that could be deduced from the normal meter of the text. Over time, however, the songs became more complex and several shapes were developed for the notes to indicate how long to hold them.
Block PrintingBy the mid 15th century, monastic musical notation was pretty well standardized. When people started using woodcuts for text, it seemed natural to start using them for music as well. The first use of woodcuts for printing music was in drawing just the staff, with the notes done in by hand, around 1442. By the 1460s, however, most musical printing was done by carving out both for the staff and the notes:
Woodcuts are difficult, and some aspects of musical notation made them especially so. The 'white' notes, or notes cut to have gaps in their centers, were troublesome when they fell on staff lines, as they were very tricky to carve. Not changing the notation system in response to technical challenges is a common theme we will see again, and probably reflects the greater volume of handwritten music.
Early Typographic PrintingThe idea of printing text with movable type led shortly to typographic musical printing. The first attempts at musical type were two pass affairs, once for the staves and again for the notes. These were very nice when carefully done, but tricky because lining up the paper for a second impression and setting type in exactly the same place in the form were both difficult.
One solution was to create type that had both the note head and a horizontal section of the staff. This sort of type could be set very quickly, much faster than double impression printing or than carving woodblocks and quickly spread. It was not nearly as elegant as other systems, and the type aged poorly. Most of its problems were due to the sections of the staff not joining up exactly to make clear lines, a problem that was not solved until the invention of stereotyping in at the turn of the 18th century. These malalignments can be seen clearly:
Typeset music was also not very adaptable. It was restricted to single lines of melody, had limited flexibility in spacing, and as musical notation was extended the notation to connect related notes horizontally, there was no way to typeset that. Bars and ties were eventually implemented with lines above or below the rest of music, but they were hard to read, being too far from the notes they modified. These restrictions kept typographic music mostly in the high volume religious vocal trade, with instrumental music left to engraving or handwriting.
EngravingIn the mid 17th century people adapted the process of copperplate engraving to the production of music. Engraving is an intaglio method, one in which the portions to be inked are removed or depressed. Printing from a plate is a matter of inking the whole plate and then wiping the ink off the non-engraved portions, so there is ink only in the indentations. The plate can then be put through a strong press with damp paper, and the paper gets precisely inked. Here we can see a major advantage of engraving where three notes (those above "Love's") are both barred to mark rhythm and indicated as slurred, marks that could not be made with type:
Because engraving requires only the black portions to be removed it is much faster to engrave plates than carve woodblocks. Woodblocks made up some of this time in being faster to print with, requiring only a surface inking and a stamping, but were hard enough to inscribe music on that once engraving became workable for music woodcut music almost entirely disappeared.
In the early 18th century a method of using pewter instead of copper for the plates began to spread. Pewter was much cheaper than copper and it also softer, so notes could be inscribed with a punch. Punches dramatically sped up the work of the engraver, who no longer had to carefully remove an oval region for a single note. With punches the creation of a plate was now no slower than typesetting the same amount of music and still much nicer looking.
Printing from plates was also nice because it could avoid the overstock problems common to typographic printers. Keeping type set for future printings would have been impractical as type was expensive and needed for other works. With a plate, on the other hand, the only expense was that of storage. This allowed engravers to ink the number they expected to sell and no more, knowing that if more were required they could simply ink additional ones.
There were still advantages to typeset music, though, and in high volume works it continued to be the standard. The main problem was that the inking and imprinting process for plates was very time consuming compared to taking an impression from type. One solution to this time problem that was not widely adopted was to ink the plate directly and make impressions from that as a normal printer would. This would put the notes and staves in white while the background would be the color of the ink, which was usually blue or green. This method was how printers normally made proofs, and they could easily create impressions in this manner, but the only attempt we know of to sell such prints, starting in the 1840s and lasting around twenty years, was a commercial failure. It was much cheaper per copy, but musicians were not willing to buy music sheets that were not black on white, refusing again to make concessions to the technology.
LithographyLithography, now used for almost all high volume reproduction, was developed for music reproduction. Invented in 1796, it did not become well known until the early 1820s, but soon was the standard. To print a lithograph the printer would, through various methods, get something greasy on a lithographic stone in the pattern they wanted to print. They'd put acid over the stone, removing the surface in non-greasy areas, then wash away the acid and grease leaving raised areas matching the pattern. Impressions taken from the stone would get ink only where the greasy substance had been.
Initially most printers wrote directly on the lithographic stone with a greasy ink. This allowed the printer to reproduce anything that could be handwritten, but it did not look as regular as music that was produced on punched plates, and could be harder to read. One solution was to create a plate, make a single print with a greasy ink, put that print on the lithographic stone, and use chemicals to get the ink to transfer to the stone. This method produced excellent results, left a plate that could be stored, and made reproductions rapid. It did require a lot of equipment, everything required for both engraving and lithography, but the other benefits were great enough that this became very common.
Mosaic-based Typographic PrintingAs the notation for music became more complex the old typographic system became less and less well suited for the reproduction of music. The idea of setting each individual notational element separately had been around for a while, but it was not until the late 19th century that type founding became accurate enough to make it practical. People began to cast musical fonts with thousands of tiny components which would be combined in a mosaic to set the piece.
Because musical type was expensive and the mosaic system required a very large amount of type, the stereotyping  process was essential to this sort of printing. In stereotyping the printer sets type as normal, but instead of making an impression they oil the type and make a plaster cast. This cast can then serve as a mold for casting a single large 'piece of type' that can be treated as ordinary type. This saves wear on the type, frees up type for use on other pages, requires less type over all, and allows for very large print runs. It also allows saving an intermediate representation for future print runs, getting some of the advantages of plate based methods.
The sterotyping process also provided a stage at which some engraving techniques could be applied, when the plaster was still wet. This was used primarily for fixing places where the staff lines did not quite meet. Other modifications could be done at this stage too, such as the addition of ties between notes if the music font did not support them. Changes at this stage were challenging because the printer had to get the depth exactly right, but it was freeing to be able to make them.
Typeset music also worked very well with text. You could include text on plates, either by engraving or punching, but it was slow and, especially if punched, tended to look uneven. For small notations on instrumental music it wasn't really a problem, but for lyrics it was troublesome. The easy mixing of music type with ordinary type helped typographic music compete with engraving, especially in songbooks.
Then along came computerised printing, which I haven't gotten into at all but has made basically all of this obselete.
 Kilmer, A. D. (1971). The Discovery of an Ancient Mesopotamian Theory of Music. Proceedings of the American Philosophical Society 115(2):131-149.
 Boethius, A. M. S. (1989). Fundamentals of Music Yale University Press, New Haven CT. Translated by Calvin M. Bower.
 Poole, H. E. (1990). Music Printing. In Music and Publishing. W. W. Norton and Company, New York.
 Krummel, D. W. (1975). English Music Printing 1553-1700. Oxford University Press, London.
 Gamble, W. (1923). Music Engraving and Printing: Historical and Technical Treatise. Sir Isaac Pittman and Sons, London.
 This process is actually where the term 'stereotype' comes from.