by Iver P. Cooper

This Appendix explains, in more detail, how I came up with the estimates for the number of books in Grantville as explained in my article, My Name Is Legion: Copying the Books of Grantville. It also includes additional information about some of the copying technologies.

[Can anyone help me get copy of Overgaauw, Fast or slow, professional or monastic. The writing speed of some late-medieval scribes, Scriptorium 49(2): 211-27 (1995)?]


Bottom-Up Approach

School Libraries

Just in the Dewey Decimal classifications 500s and 600s, the middle school library has almost 900 circulating books (based on my photos of the Mannington Middle School Library). The grand total is probably at least four times that, judging from my picture of the entire room. I have never seen the high school or elementary school libraries, so I don’t have an estimate for them.

The National Center for Education Statistics says that in 1999-2000, the average number of books per pupil in public school libraries was 17. (NCES 2005-324) I can confirm the reasonableness of this number from another source. According to the U.S. National Commission on Libraries and Information Science, in 1993-94, “school library media centers” (elementary, junior and senior high, whether public or private) held a total of 879,000,000 volumes. (NCLIS) In 2000, there were 50 million students, which implies 17.6 books per student. (I realize I am combining numbers from different time periods, but only for confirmation).

According to the West Virginia Department of Education, the 2nd month 2001-2 enrollment data was as follows:

North Marion High School (G9-12): 943

Mannington Middle School (G5-8): 390

Blackshere Elementary School (K-4): 437

I am assuming the enrollment in 1999-2000 wasn’t much different. So if we multiply each of these numbers by 17, we get 16031, 6630, and 7429, respectively.

We can check these predictions against categorical data in another NCES report. According to the ELS:2002 study (NCES 2005-302), in 2001, 38.8% of “School Library Media Centers” (SLMCs) had fewer than 8000 books (volumes), 36.1% had 8-16,000, 18.9% had 16-24,000, and 6.2% had more. (Table 4). (These figures consider both public and private schools.)

As you might guess, the library size was a function of school size. The elementary and middle schools are on the borderline of ELS:2002’s 1-399 and 400-799 categories. For NMHS, the relevant SLMC category seems to be 800-1199 enrollment. Location does matter, but not as much. The ELS:2002 breakdown for the relevant categories is

ELS:2002School Library Holdings, by selected school characteristics
School Size (enrollment)
All Schools38.6%36.1%18.9%6.2%

NCES report 2004-313 gives the following interesting figures for average number of books held at end of 1998-99 by public school libraries:

West Virginia: 6,873

Rural/Small Town: 9,125

Elementary School: 9,375 (this category includes middle schools)

Secondary School: 13,164

200-499 students: 8,583

750-999 students: 12,886

All public school libraries: 7,192

For right now, I would call it 8,000 apiece for the elementary and middle schools, and 13,000 for the high school.

Public Library

According to the West Virginia Library Commission Statistical Report 2005, the Mannington branch of the Marion County Public Library had 23,707 print materials, 1,111 audio materials, and 3,087 video materials. Some, of course, will be on loan to people outside the RoF. By the same token, people in the RoF may have borrowed materials from the Fairview branch, or the central library in Fairmont. I figure it balances out.

The standard (NCES) meaning of “print materials” is the total of the number of books and the number of bound serial volumes. We don’t need to exclude the latter since the downtimers would want to copy them, too. However, just to give you a sense of the ratio, in 2000 the Maryland public library system had about 14,000,000 catalogued book volumes, about 653,000 uncatalogued ones, and about 692,000 serial volumes. (MdPubLib)


Place of Business Libraries

This will include the books that the company owns (e.g., a maintenance manual at an auto dealer, or the West Virginia statutes in a law office) and those personal books which its employees keep at the office. The Uptimers’ Grid lists about a hundred pre-RoF businesses. Some are likely to have only a few books (stone monument service), others, a significant collection (Dr. Adams). I studied the pet shop my wife frequents and they had at least twenty five books, not counting catalogues.

Based on the Grid, I have compiled the following list of pre-RoF businesses which survived the RoF:

The ones which I am fairly sure existed pre-RoF are

*medical offices, MD (Adams)

*medical offices, DO (Shipley)


*dentist (Sims)


*pharmacies (Nobili’s, Moss and Little’s, Trelli’s)


*churches (Catholic, Methodist, Presbyterian, Southern Baptist, Church of Christ, Disciples of Christ)

machine shops (Nat Davis, Ollie Reardon, Dave Marcantonio)

grocery stores (Stevenson’s Groceries, Garrett’s Super Market, Johnson’s Grocery, Quik Mart franchise)

*thrift shops (Dollar Store, Bargain Central, Bits ‘n’ Pieces)

*sporting goods stores (Bartolli’s Surplus and Outdoor Supplies, Grantville Guns and Tackle, Grantville Sporting Goods)

hardware store

Clarence’s Heating, Plumbing and Air Conditioning

O’Keefe’s Septic Tank Maintenance Company

**Grantville Times

*Jewelry Store (Morris Roth)

*Bridal Shop (Karen Reading)

Restaurants (250 Club, McDonald’s, Castalanni Brothers, McDougal’s Restaurant, sandwich shop next to City Hall, Tyler’s Restaurant, Casey’s, Marcantonio’s Pizza, Mora’s Café, Steak ‘n’ Shake drive-in, Tip’s)


*Gift Shop

*Old Barn Museum

*Historical Society Museum

*Video Rental store

*Tack Boutique

*general contractors (Happy Acres, Home center)

home repair and handyman firm

lumber company (Home Center, see above)


beauty salons (Carole’s, The Candy Cane, Eileen’s, Curl and Tan)

laundries/laundromats (Rawls’ MaidenFresh Laundromat, Morton’s Self-Service Laundromat, Mi Casa Laundromat and Self Service Cleaning)

funeral homes (Central, Genucci’s)

stone monument service

day care centers (A&A, Toddler Haven)

storage rentals (Buster’s, Higgins, Clyde’s)

**banking (Bank of Grantville, Grantville Credit Union)

**accounting firm

**real estate agents (Grantville Homes and Land, Lamb Commercial)

*elder care centers (Bowers, Manning, Prichard)

We also have the following former businesses listed in the Grid:

gas stations


auto repair and body shop

mobile home moving

*car dealers (Trumble, Jenkins, Barlow, Lowe)

* personal/home office grade photocopier (8 ppm) predicted.

** office grade photocopier (20+ ppm) predicted. The schools and the town government will also have them.


Home Libraries

A scholarly paper (Kelley) which used the number of books in the parent’s home as a measure of parent’s scholarly culture. For the US it reported an average of 162. Note that there was some statistical bias in that the max choice was “over 500” and that for averaging purposes that was treated as 600.

To get a home library size per person, we would have to divide by the number of members of the household. The average size for a family household (Statistical Abstracts table 55) was 2.62. So that yields an average of 62 books per person. There are 3500 residents in Grantville, so that’s 217,000 books.

That estimate is based, unfortunately, (1) on nationwide data rather than data for Mannington, Marion County, or West Virginia, and (2) studies of households with children rather than all households. Only 28.9% of Grantville households include a child (Census 2000 for Mannington).

NCES Table 20-1 has some interesting information on adult reading habits in 2003, e.g., a breakdown of the percentage of adults aged 16 or older, by sex, age, education, race, and household income, with respect to how often they read books and whether they have 25 or more books at home. The national average for the latter is 88.2%. The groups with a rate less tan 80% were age 19-24 (79.6%), less than high school education (72.3%), Hispanic (66.9%), and household income less than $15,000 (71.7%). I applied the Grantville (Mannington 2000 Census) education demographics to the NCES data and got a predicted rate for Grantville of 85.8%.

If acquiring books is a Poisson process, then if 85.8% have 25 or more books, the average person has about 30.5. However, one of the assumptions for a Poisson process is that the occurrences are independent, and I think that obtaining one book makes it more likely to buy another.

Distinct Titles

Distinct titles, as a percentage of total holdings, for various libraries ranged from 18 to 83%. The weighted average was around 25%.

I have looked at data for several public and school libraries, and consortiums of libraries (see Appendix), and I think that the data follows a rough power law. That is, the ratio of the logarithm of the total number of volumes to the logarithm of the number of distinct titles seems to fall mostly in the range of 1.05-1.15. That means that the number of distinct titles, expressed as a percentage of the total holdings, decreases as the library size increases.

If I use the ratio 1.1, then if the total holdings of Grantville are 280,000, the number of distinct titles in Grantville is 90,000. With the more conservative ratio of 1.15, the distinct titles drops to 55,000. If I keep the ratio at 1.1, but use the more conservative holdings estimate of 161,000, I get 54,000 distinct titles.

Number of Books in Grantville: A Top-Down Approach

I am going to try to come up with the total number of books in Grantville in a different way. Let me start with information from the 2006 Statistical Abstract of the United States. Table 1125 Quantity of Books Sold says that the net publisher sales (by publishers in US, and imports under US imprints) in 2000 was 2,461.9 million books. Those books, of course, would find their way into libraries, schools, offices and homes. The population of the US in 2000 was 282,192,000, per table 2, and the median age was 35.2 years, per table 11.

So, if we assume that the level of purchases per capita per year has been constant over many years, and that no books, once acquired, are lost or destroyed (selling or giving a book is just a redistribution, doesn’t change the total) , then the average books per person is (net sales / population) * median age. The per capita net purchases in 2000 were 8.7 books. Multiplying by the median age yields an estimate of 307 books per person. Remember, that number includes not only the home library, but also each person’s pro rata share of the books owned by institutions. (Yes, I realize that one year olds aren’t buying books. The use of the median age is still proper because it accounts for the middle-aged and older folks who buy books.)

Unfortunately, I have not been able to do much of a test of the first assumption. In 2005, books sold was 3,078.9 million, and the population was 296,639,000. That is a per capita of 10.4 books.

In Grantville, the 2000 median age is actually higher (40.47 years). 40.47*8.7=352/

However, we have reason to believe that reading activity, and hence book ownership, is lower in Grantville than nationwide.

For example, only 12% of residents 25 and older have a college degree. The national figure (Table 214) is 25.6% (the West Virginia average is 15.3%). If we correct the annual purchase estimate by assuming that it is proportional to the number of residents with college degrees, we get 4.1 new books per person per year. And, if we use the Grantville median age, and again ignore discards, we get 165 books per person.

If we use the per capita library circulation ratio (2.7/6.4=0.42), we have 3.65 new books per person per year, and we get 148 books/person.

While there are no statistics on how quickly ordinary citizens discard books, libraries do compile that information. For example, in the Pennsylvania School Library System, in 1998-99, 5340 books were added, 1,867 were withdrawn, and the holdings at the end of year were 990,672 volumes (403,420 titles). That’s a discard rate of about 0.2%. The rate in 1989-90 was more like 1%. For the Marshalltown public library, 75,199 books were held at start of 2004, 5,780 were added, and 5,155 withdrawn. That’s a discard rate of 6.8%.

(Sources: ; )

Again, bear in mind that a discard by one uptimer may end up, through gift or sale, as an acquisition by another. So the effective discard rate — the rate at which books are truly lost from the Grantville “pool” — is less than the nominal rate.

I created a simple spreadsheet model. Each year, I subtracted the discards from the model person’s holdings, then added 4.1 new books (the education-based number). For a discard rate of 5%, the person had 71 books at age 40. For a rate of 1%, the person was left with 136 books. If I do this with the library circulation based number (3.65/year), I get 64 (5% discards) or 122 (1% discard).

Distinct Titles

I am deliberately using the term distinct titles, instead of the more usual unique titles, because consortiums use unique titles to refer to titles held by only one member library. What I am interested in is the total number of titles available in at least one copy in the library or libraries in question.

Relevant public library data is available — the Normative Data Project for Libraries’ circulation measures — but not available for free. Nor is there statewide info for WV. The fact is, a lot of libraries have given up trying to track the number of distinct ttles.

I have found a smattering of relevant information on the internet.

Table: Distinct titles as percentage of number of items

Illustrative Public Libraries

Dekalb County (Decatur GA): 32%

Texas (statewide): 64%

Santa Fe: 61%

Cleveland: 25%

Santa Cruz City-County: 53%

Canton (Michigan): 83%

Illustrative School Libraries

Pennsylvania School System: 41%

[citations omitted]

I have theorized that the relationship of distinct titles to total holdings isn’t normally distributed around a mean percentage, but rather follows a power law. The reason I suggest this is that the popularity of books follows a power law. If the power law distribution is right, then log titles = INTERCEPT + SLOPE * log holdings. The result I got was INTERCEPT=0.306, SLOPE=0.870.


Words per Book

The books in Grantville are diverse, including children’s stories, textbooks, novels, “how-tos”, and encyclopedia volumes, which vary greatly in word count. Children’s books average 1000 words. The 1992 edition of the classic Morrison and Boyd textbook on Organic Chemistry is almost 363,000 words. Eric Flint’s 1632 is 173,787 words.

A rather informal study (Crawford) of the Amazon book database suggested that the average book had 70,000 words. The same figure appears in the Association of Book Publishers of British Columbia’s Non-Fiction Proposal Guidelines. The characters in Oskar’s The Hand of Fate (p. 22) say that the average book (based on a sampling of bookstore paperbacks) was about 100,000 words. That number also appears in Rasula, Imagining Language: An Anthology 31.

So let’s say 70-100,000 words.

For some purposes, it is helpful to know the number of words per printed page. For average-size adult books, this is typically in the 300-600 range, depending on the font and margins. For EB1911, it about 1500.


Hand Copying

1. In late medieval society, university stationers rented out small text selections by the week. The selections were eight pages (16 columns) long. A student could copy three or four columns a day (seven in a rush). (Schwarz 216). Unfortunately, I don’t know the number of words to the column.

2. Writing speed is influenced by whether you are using block letters or cursive, whether you are an occasional copyist or an experienced scribe, and whether you understand the language of the writing you are trying to copy. Cursive writing, by the way, is not common in the early seventeenth century (see “Cursive,” Wikipedia). Cursive is difficult for others to read and isn’t likely to be used for anything more than personal notes.

3. Longhand copying speed could be as high as 25-30 wpm (Schwarz 226).

4. For some of the works, there will be economies of scale which will lead to those being printed on a letterpress. That doesn’t mean, however, that there might not be an intermediate hand written copy. For example, if the work is being printed in Bamberg, then either the original must be bought/borrowed, or a copy must be made to be sent to Bamberg. Also, there are going to be works which interest too few people to make it into the printing avenue. So those might be transmitted as handwritten copies.

5. A late nineteenth century race between a typewriter and a penman involved copying of 1,000 words of statute law. The typewriter took 19.5 minutes (speed 51 wpm), whereas the scribe required 108 minutes (9 wpm)(LaFollette 369). That seems unduly low.

6. The results of twentieth century handwriting tests have been published, but these have their own problems. They are usually administered to children, not adults, and sometimes the children are learning-disabled. Also, the test protocol is not always ideal. What we would prefer is to have the subject write out as much as possible of a long text, under a time limit. However, an alternative protocol is to have a word or short phrase written as many times as possible within a set period of time.

7. Twentieth century estimates implicitly assume the availability of modern pens and paper. With seventeenth century materials, hand copying could be slower. A scribe in action would have to continually re-dip the quill as the ink on the tip ran out. And as the quills wore out, they had to be re-cut or replaced.


The principal value of shorthand is for taking dictation, not for copying written texts. Shorthand is faster than longhand, and shorthand systems existed in the early seventeenth century, but they are immediately relevant only to those who are making a copy for personal use. The problem is that the systems weren’t sufficiently standardized so one person could expect to be able to read another’s shorthand and, more importantly, the percentage of even the literate population who understood any shorthand was small. The people who know shorthand are the professional scribes, but the folks who are paying for copies to be made are nobles, merchants and craftspeople.

It is likely that up-time shorthand systems will become known to downtimers. Those who take the trouble to master them can achieve high speeds (the manual shorthand record is 350 wpm with the Pitman system during a two-minute test by Nathan Behrin in 1922, according to the Guinness Book of Records).


1. It is important to bear in mind that since the typewriter surveys were conducted by email or at a website, all the respondents, by definition, own computers. They therefore might not be representative of the Grantville population, in which computer ownership is probably down in the 20-30% range. However, you could argue that a person with a computer is more likely to toss out an old typewriter than one without (I junked my Smith-Corona a decade ago.)

But if I cannot determine the total # of people in the Baen’s Bar households because several people didn’t answer that question.

2. The first typewriter known to have been assembled after the RoF was the one mentioned by Huff, Other People’s Money (GG3): “after a six-week intensive course he [David Bartley’s down-timer secretary] could take shorthand and was learning to type on a custom-made typewriter. ” From context, this occurred before the Croat Raid, which is mentioned later in the story. In view of 1634: The Galileo Affair, we have to assume that “custom-made” means assembled from several non-working up-time typewriters.)

3. In 1907, average typing speed for a typist was 50-60 wpm; a “world champion” could type 87-95 wpm. (Schwarz 226).

4. Wikipedia, ‘Typewriter” says, “As of 2005, Barbara Blackburn is the fastest typist in the world, according to The Guinness Book of World Records. Using the Dvorak Simplified Keyboard, she has maintained 150 word/min for 50 min and 170 word/min for shorter periods and has been clocked at a peak speed of 212 word/min.”

5. There is a lot of disagreement as to how long it takes to learn “touch typing” (or keyboarding, as it’s called nowadays). There are computer programs which claim that you will be able to touch type after just ten hours of practice (although they tend to be rather coy as to how fast you’ll be).

According to the website of a British commercial college, their “copy typing course (elementary exam level),” requiring sixty classroom hours, would prepare you to touch type at 25-30 wpm, and get a job as a clerk-typist. If you wanted to be hired as a secretary, you were encouraged to also take their intermediate typing course, requiring another sixty hours.


Then there is the hectograph (“jellygraph”). In essence, you write (or type) on the master with a special ink (an aniline dye), and press the inked side down on a gelatin pad. You then apply the copy paper to the same pad, and the image is transferred. This method is good for only twenty to eighty copies. So “hectograph” is something of a misnomer (hecto=100). Hectography is also slow, as it takes several minutes for the ink to transfer properly.


With my Epson Perfection 2400 Photo, set at 300 dpi, I spent four seconds laying and weighing down the book, five seconds running the preview scan and setting the margins accordingly, and eleven seconds running the final scan. That’s an effective speed of three scans/minute.

Project Gutenberg FAQ reports 20-40 seconds for one scan at 400 dpi, so about two scans/minute.

U. Michigan says (p. 5), “Material that is scanned in bound form often requires a great deal of post-scanning enhancement in order to flatten out the curvature of the volume and to remove gutter shadow.” They were able to disbind at a rate of 30 volumes/hour.

Speech Recognition

In 2006, recognition of speech at 160 wpm was possible (DMT), but required a fast computer with lots of RAM (DMT). There are perhaps forty Pentium IIIs and twenty Power Macs in Grantville (Bartholemy). The reader would almost certainly need to be an uptimer with good diction.


Daisy Wheel Printers

A typical daisy wheel printer was the Xerox Diablo 630 (1980); speed 30 characters per second (cps) or 300 wpm.


Dot Matrix Printers

This will probably be the first computer printer technology to be re-invented in the 1632 Universe. Compared to a manual typewriter, a DMP is fairly simple; in 1980 the average DMP had 150 parts; the IBM ProPrinter had just 62 (Utterback 142). It is uncertain whether down-time made DMPs will have the same graphics resolutions as uptime machines.

We may also see hybrid typewriters which combine a typewriter input and a DMP output. Such machines were marketed in the 1980s. (Wikipedia, “Typewriter”).

With a 9-pin DMP, the pins form a 9×1 matrix; a pin spacing of 0.35 mm results in an effective print resolution (for graphics) of 72 dots per inch (dpi). With a 24 pin DMP, the pins form a 12×2 matrix, and a pin spacing of 0.21 mm yields 120 dpi per column, or 240 dpi overall. (Mimech)

The 9-pin Epson FX-880 was released in late 2000, and therefore is probably a bit superior in speed to any dot matrix printer available in RoF Grantville. It had a maximum draft speed of 410-455 characters per second, a normal draft speed of 310 cps, and a “near letter quality” mode of 77 cps. The NLQ mode is equivalent to about 720 wpm (with 300 words/page, that’s 2.4 pages/minute). For graphics, you would use NLQ mode; resolution was 240 x 144 dpi (for best results, printing unidirectionally, which would halve the speed). The contemporary 24-pin Epson LQ-2180 had high speed draft of 480 cps, normal draft 360 cps, and NLQ at 120 cps (1200 wpm). Graphics resolution was 360×360 dpi.

For the FX-880, the life expectancy of the printhead was 200 million characters, and the ribbon was good for 3 million in draft mode. Print volume of 6.5 million lines, mean time between failures of 10,000 printer-on hours. For the LQ-2180, the printhead had a 400 million character life expectancy, and the ribbon, 8 million. Print volume 24 million lines; same MTBF as the FX-880. Bottom line is that barring catastrophe, any DMP which is operable as of the RoF will continue to operate for years thereafter. So even if we have trouble building DMPs downtime, it isn’t a big problem.

Laser Printers

The first mass market laser printer was the HP Laserjet (1984), which had a speed of 8 ppm and 300 dpi resolution. If that was printing a typical double-spaced, courier font page, that works out to 2,000 wpm. Later, HP produced markets for different segments. For example, the HP Laserjet IIIP (1991), for the personal printer market, had a speed of only 4 ppm. Office printers included the HP Laserjet 4Si (1993), 17 ppm, and the Laserjet 5Si (Nov. 1995), 24 ppm. The laser printers of the late Nineties typically offered higher resolution (up to 1200 dpi), but not faster speeds. (HP)

Inkjet Printers

The first successful commercial inkjet printer was the HP Deskjet (1988)(2 ppm, 300 dpi, monochrome). The HP DeskJet 882C (Jan. 1999) was rated at 5 ppm color, 8 ppm black, 600 dpi.


1. I have tabulated the speeds of selected pre-RoF Xerox monochrome copier models:

4000 (1970) 45 copies/minute

9200 (1974): 120 (high volume)

1090 (1985) 92 (reduces/enlarges)

1065 (1987) 62 (mid-volume, full-featured)

5042 (1989) 35 (mid-volume book copier)

5100 (1991): 100

4135 (1991): 135 (high-volume)

5209 (1991): 8 (small/home office)

5126 (1993) 26 (book copier)

5800 (1998): 120 (“departmental copier”)

DocuPrint 180 (1998) 180

Xerox 5890 (1998): 100 (“compact high volume”)

Bookmark 21 Copy Station (1999) 21 (coin- and card-operated models)

(Xerox Factbook)

Bear in mind that the quoted copy speeds always assume ideal conditions. That is, making multiple copies of a single, cut sheet original, without pauses to change paper or add toner.

2. A photocopier with a rated speed of 22-32 copies/minute might be expected, with normal maintenance, to have a life of 1,000,000 copies (Visualhawk). You’re not going to get anywhere near that, because you’ll run out of paper or toner first. And even if you didn’t, sooner or later there would be another problem requiring spare parts or expertise which no one in Grantville has.


Books were sold unbound, so the buyer could choose a fancy or a plain binding, as his or her purse allowed. Sometimes, there were two versions, a big run on paper, and a small one on vellum for the well-heeled customers. The “premium” version might also have fancy artwork; an artist might charge 60 florins to decorate a single volume (148).

The Book Market

Average retail prices for books in 1998 were $49.60 (hardcover), $22.86 (trade paper), and $9.31 (mass market paper)(Statistical Abstracts 2001, Table 929), and average hourly wages were then $12.78 (Table 682), and federal minimum wage $5.15.

In seventeenth century Britain, the price of a “typical” book was 1.5-2 shillings ((Maxted); a laborer was paid about eight pence (2/3 shilling).


Down-Time Book Ownership

Hall, “The Old and the New: Literacy and Reading in Eighteenth Century America,” page 16 .

McCraig, English Lute Manuscripts and Scribes 1530-1630, Chap. 4,

Maxted, “A history of the book in Devon. 35: Books and readers in the 17th century,” (2001),

School and Public Libraries

NCLIS, “Basic Summary Statistics of US Libraries,”

NCES 2005-324, “America’s Public School Libraries; 1953-2000,” Table 1,

NCES 2002-308, “Public Libraries in the United States: Fiscal Year 1999,” Tables 4, 4A, C2, C3 (2002),

NCES 2002-344. “Public Libraries in the United States: Fiscal Year 2000,” same tables,

West Virginia Dept. Education, “WV Achieves Report Card,”

NCES 2005-302, “School Library Media Centres: Selected Results from the Education Longidtudinal Study of 2002 (ELS:2002),

NCES 2004-313, The Status of Public and Private School Library Media Centers in the United States: 1999-2000, Table 6a,

West Virginia Library Commission, “Statistical Report 2005,” page 20

MdPubLib, “Maryland Public Library Statistics,”

Home Libraries

[NCES] National Center for Education Statistics, Table 20-1, in “The Condition of Education” (2006),

(2003 data)

Book and Page Word Counts


Crawford, “Losing What Counts: The Swamping Phenomenon”

Hand Copying


Allcock, “Handwriting Speed Assessment, Update September 2001, Testing Handwriting Speed,” in PATOSS (Professional Association of Teachers and Students with Specific Learning Difficulties),

Hilz, The Network Nation: Human Communication Via Computer

Topik, The World That Trade Created: society, culture, and the world economy, 1400-the present

Bishop, “Writing Speed and Extra Time in Examinations,” 5th BDA Intl. Conf. (2001),



LaFollette, The Making of America (1906)

Ostrach, “Typing Speed: How Fast is Average,

Frank, “Tower Clock Collection, Page One,”

Groover, Automation, production systems, and computer-integrated manufacturing (19__)

Cipolla, Clocks and Culture: 1300-1700 (19__)

Ayres, Computer Integrated Manufacturing (1991)

Utterback, Mastering the Dynamics of Innovation (1996)

Davis, “Typewriter Designs: Key Lever and Type Bar Mechanisms,”

Loveday, Job-entry typewriting speeds of three different levels of secretaries ar a large public university,”

(mean 61 wpm, range 28-99).

“Civil Service Exams,” (40 wpm)


“Profile Portraiture: A Techno-Art Interlude,”

“Jefferson and the Polygraph,”

Spirit Duplicators, Mimeographs, Hectographs

Wikipedia entries for above

“A.B. Dick Company”, in Encyclopedia of Company Histories,


Xerox FactBook 2003-2004, “Major Products by Year”,


Monash, “Purchasing photocopiers. Size and speed issues: Matching copy volumes to copier speeds,”


Bartholemy et al, “Computers in Mannington

Scanning and OCR

Adams, “How to Scan a Book,”

ComputerWorld, “Optical Character Recognition,” (July 29, 2002),,10801,73023,00.html

Rice, et al., “The Fourth Annual Test of OCR Accuracy” (1995),

Cohen, “How to Make Text Digital: Scanning, OCR, and Typing,”

Project Gutenberg, Scanning FAQ,

User Review (2003) , “OmniPage Pro 12.0 Office,”

(reported that with 700 page job, program took two hours to load, 8 hours for OCR, and 2 hours to save RTF file. That’s about 60 pph).

U. Michigan, “Assessing the Costs of Conversion: Making of America IV: The American Voice 1850-1876” (2001), (Collating (check for missing pages), 3 vols/hr, 2515 hrs. This was for MoA4, almost 8000 vols, >2,500,000 pages. 600 dpi scans. OCR — with expensive software (PrimeOCR, >$10K)– 99.8% accurate. says 7576 volumes, 2347044 pages of content.

Bicknese, “Measuring the Accuracy of the OCR in the Making of America” (1998),

Shaw, “OCR and SGML Mark-up of Documents from the Making of America Project” (1996)(used ScanWorx, one CD-ROM, 4000 pp, 7-10 hrs for initial OCR processing with scripts + 1-2 hrs staff time to handle exceptions, so 333-500 pph for OCR. Elsewhere says 650,000 pages in project, 2 sec/page for auto OCR conversion (inconsistent!), 2-3 min/page for SGML markup, 8-9 min/ page for proofreading and correction).

Henschen, “OCR for Business: Beyond One Desktop,” Transform (Feb. 2004)(Omnipage Pro 14 processed 20 pages at rate of 140 pph; FineReader at 343 pph. Accuracies ~99.5%. Note that these are post-RoF versions).

Speech Recognition

Devine, et al., “Comparative Evaluation of Three Continuous Speech Recognition Software Packages in the Generation of Medical Reports,” J Am Med Inform Assoc. 2000 Sep–Oct; 7(5): 462–468,

Kramer, “Speech-Recognition Technology for Computers,” Acad. Pyschiatry 23(1): 48 (Spring 1999), Link via

[DMT] “History of Speech & Voice Recognition and Transcription Software,”

Johnson, Update on Voice Recognition: Will it work for you?” (WATA Bulletin, Fall 1999),

Griffith, Speech Recognition for Ijury, Disability and Prevention (1998),

Patterson, “Dictation Software: we’re not there yet”, JAMC 160(6): 886 (Mar. 1999)


Wald, M. (2005) “Automatic Speech Recognition Assisting and Enabling Receptive Communication. In Proceedings of Balancing the Equation: Key issues in learner support in HE from the Assistive Technology perspective,”

[WikiWPM] Wikipedia, “Words per minute,”


Bailey, “Human Interaction Speeds,”

Weiler, Email (Oct. 27, 2003), in “A significant OCR Milestone, Book People Archive,,17

Hah, “Comparison of Speech with Keyboard and Mouse as the Text Entry Method,” Proc. Human Factors Ergonomics Soc’y 49th Ann. Meeting 619 (2005),

Computer Systems

Census Bureau, “Home computers and internet use in the United States” P23-207 (2001).

Infotrends, “Scanner Population Continues to Grow…,”

Kraemer, “Printing Enters the Jet Age,”

[HP], “Twenty Years of Innovation: HP LaserJet and Inkjet Printers 1984-2004,”

Epson printer product brochures.

Typesetting; Printing

Sonn, Paradigms Lost: the life and deaths of the printed word (2006)

Jardine, Worldly Goods (1996)



Proz: The Translators Workplace, “What is the realistic translation speed,”


Schwartz, The Culture of the Copy (1996)., “Antique Copy Machines,”