Carving stone tablets at home in your spare time

[Kragen Sitaker]

(Haven't tried this yet.)

At dinner the other night, I was discussing information storage with a
bunch of other people.  Someone mentioned that he had been to a
librarians' conference at which some librarian made the assertion that
acid-free paper was the only "reasonably long-term" archival medium we
have.  I demurred and described the technology used by the Rosetta
Project: a micro-engraved nickel disk.

Someone else suggested that perhaps the librarian had overlooked stone
tablets as a medium of information storage, but apparently this
librarian had not considered stone tablets practical.

At this moment, it occurred to me that stone tablets had recently
become practical for the everyday American, who can use the following
method, which I have not myself tried, but which I am pretty sure will
work.

1. Print out a one-page document on a laser printer, backwards
and, ideally, with black and white interchanged.  (There's a little
bit of PostScript magic that can achieve this.)  
2. Place the paper face down on a polished slab of marble.  
3. Soak the paper with a solvent such as Citra-Solv.
4. After it's been soaked for a little while (a few seconds to a
minute should do), peel the paper away from the stone, leaving some of
the toner stuck to the stone, backwards --- forming a perfect image of
the original document.
5. After the solvent has completely evaporated, etch the stone lightly
with acid --- I think hydrochloric acid is the traditional choice for
this, but Coca-Cola should work fine, just slower.  The acid will
not etch the toner.
6. Rinse the stone with water to wash away the acid.  

Now you have an etched stone tablet.

There are several potential limitations on this process:

- The acid might etch the toner.  As I wrote this, I tested with some
  laser-printed paper in a glass of McDonald's Coca-Cola, which
  unfortunately came with ice, for about 20 minutes.  The paper was
  pretty fragile afterwards, but the letters on the surface did not
  smear and still retained some structural integrity when the paper
  beneath them tore.  So I don't think the acid will etch the toner
  --- at least, not cold Coca-Cola, which I'm pretty sure will etch
  marble.
- Laser printers cannot print very small.  The most-widely-cited
  resolution I've seen lately is 1200 dpi, although older laser
  printers commonly printed only 150-300 dpi.  The larger-grained
  toners used in the lower-resolution printers produced thicker
  deposits of toner on the paper and may therefore be better suited to
  this process.
- 5x7-pixel monospaced bitmap fonts are still quite readable, so 1200
  dpi means that a square inch is 171 0.42-point-high lines of 240
  columns (41k characters or 4500 words), which is normally about 9-10
  English words per 70-character-wide line; proportional fonts are
  still readable at 7 pixels high, but take up less space, allowing
  10-14 English words per same-sized line --- or 5000-7000 words per
  square inch.  At 150 dpi, you might have only 70-110 words per
  square inch.  Optimistically, this means 654 500 words per page
  (30-100 hours of reading); pessimistically, 6200.  (Of course, you
  may prefer to print something other than tiny text; this is just
  about the limitations of text size.)  (By comparison, the
  microfiches I used to use for library catalog lookups were about
  4"x6" and had on the order of 100-200 lines of text per inch.)
- Dissolving the toner may cause it to flow and cover a larger area,
  reducing the effective resolution.
- The etching must be thorough enough that there is a visible
  difference between the etched areas and the unetched areas.  But the
  depth of the etching probably cannot be much less than the feature
  size of the image --- you won't get 1200 dpi resolution unless you
  etch the stone less than 1/1200 inch deep, and you won't get 100 dpi
  resolution unless you etch the stone less than 1/100 inch deep.
  Etching the letters instead of the space around them (applying toner
  over most of the stone) may mitigate this problem, but may have
  other disadvantages.  (1/1200" doesn't sound very deep, but it's
  still 700 wavelengths of blue light, so it's plenty deep to make the
  stone not shiny.)
- The etching won't be perfectly uniform across the stone, so to make
  your etching less than 1/100" deep everywhere, you might have to
  make it less than 1/50" deep somewhere.
- The toner may not transfer completely --- some areas of the stone
  may remain unprotected from the acid.  For this reason,
  "reverse-video" printing may not work
- Halftoning patterns will work better if the halftone screen has no
  features (dots or spaces) smaller than the etch depth.  If the
  features are very close in size to the etch depth, they may just
  translate into different etch depths.
- You may want to remove toner when you're done, leaving only polished
  marble.  This might be difficult; this process will probably have to
  be partly mechanical, rather than purely chemical.  You could try
  scraping the toner off when it's completely hard, using some kind of
  a blade, but this will be hard to do without damaging the stone.
  But if you try to wash it off with a solvent, or scrape it off after
  softening it with a solvent, it may be hard to keep the dissolved
  toner from soaking into the etched areas of the stone, which will
  have very rough surfaces.  Perhaps you could just abrade the toner
  off, but this requires either that you cease to abrade after getting
  through all the toner but before erasing the whole stone (which also
  means that your polishing surface must be flat to within the depth
  of your etching, such as 1/1200"), or that you use an abrasive
  softer than the stone itself, yet harder than the toner.  I'm not
  sure that laser printer toner is actually softer than marble.
  Keratin, talc, and gypsum all come to mind as softer than marble,
  but none is particularly promising as an abrasive.  Ground-up
  fingernail clippings or gypsum sand, mixed with water?  Sandblasting
  with gypsum sand?
- Marble tablets etched 1/1200" deep still aren't as durable as
  acid-free paper with ink under many practical circumstances.

Other thoughts: 

- Flat Coca-Cola will probably etch slower but provide more even
  etching than fizzy Coca-Cola.
- Vinegar would probably work for etching too, and is cheaper, but
  Coca-Cola smells better and is usually more acidic.  (Vinegars
  vary.)
- Other stones might work also, but marble is particularly easy to
  etch.  However, marble is weak, and so will probably need to be at
  least 1/4 inch thick, which means an A-size stone tablet will be 383
  cubic centimeters; my vague memory of marble's density suggests that
  this puts its mass somewhere around 3.5-4 kg.  A stronger and harder
  substance could be considerably lighter and easier to polish, but
  might require a more exotic process to etch, using hydrofluoric acid
  (all the harder substances that occur to me are quartz-based), or
  perhaps sandblasting with quartz sand would work.
- How big is text?  I looked at RFC 3117 (about BXXP, I mean BEEP),
  which I happened to have handy.  Pages 3-23 of this document contain
  21 pages, 717 unique lines and 6451 words (other than in the page
  headers and footers, which contain 294 words).  This gives an
  average of 307 words per page, or 9 words per line.  If you were to
  use the 5x7 monospace font mentioned earlier and print each page in
  66 lines by 80 columns as usual, each page would be 400x462 pixels,
  or (at 1200dpi) 0.385" by 0.333", and you'd fit 7.8 pages per square
  inch, so you'd need 2.7 square inches for these 21 pages.  If you
  printed the text in a proportional font that fit 14 words per 400x7
  line, omitting the blank lines and page headers and footers, you'd
  need 461 lines, or 400x3227 pixels, fitting all the text in 0.9
  square inches instead, for 25 pages per square inch.
- Some metals would work too.  Many metals are strong (much sturdier
  than marble), can be etched with acid, but stable over long periods
  of time, either exposed to air or coated with a thin layer of
  lacquer, oil, or wax.  I think copper and some steels would work,
  tin or zinc might work, and perhaps nickel, cobalt, or chromium
  (although I don't know how if they're easy to etch.)
- If you shine a bright light directionally on a partly-polished,
  partly-etched surface (for example, by putting it in sunlight),
  you'll get specular reflection from the polished part and diffuse
  reflection from the etched part.  If you look at the item from some
  distance away in the direction of the specular reflection, the
  polished part can appear dramatically brighter than the etched part.
  A lens focused on the partly-polished surface can project a
  magnified image of it on a screen.  This is particularly promising
  with the metals, which are much more reflective than, say, marble.
  Such a technique, similar to a microfilm reader, could allow many
  people at once to read some magnified text.
- A simpler, but more dangerous, magnification system consists of a
  sunlight-gathering lens focused on a pinhole at the end of a long
  cone, which functions as a point-source, letting light into a
  darkened room.  The pinhole is placed close to the tablet and moved
  around in order to project different parts of the tablet, magnified,
  on an opposite wall.  If the pinhole is 1 cm from the tablet and the
  wall is 3 m from the tablet, the magnification will be 300x.  I am
  not sure this will have sufficient contrast to read, since the
  etched areas of the tablet (and, in the case of marble, the whole
  tablet) will reflect diffusely, raising the background light level.
  It is necessary to shield the pinhole from people's eyes, because it
  will be brighter than the sun.
- This system can be made somewhat more practical and safer by using a
  black marble tablet; by using a metal tablet; by filtering the light
  that enters the light-gathering lens to exclude infrared,
  ultraviolet, blue, and possibly red light, leaving only the yellow
  light that we see best; and perhaps by stroboscopically unshuttering
  the lens at intervals shorter than the human persistence of vision.