© 2025 The Interpreter Foundation. A 501(c)(3) nonprofit organization.
All content by The Interpreter Foundation, unless otherwise specified, is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Permissions beyond the scope of this license may be available here.
Interpreter Foundation is not owned, controlled by or affiliated with The Church of Jesus Christ of Latter-day Saints. All research and opinions provided on this site are the sole responsibility of their respective authors, and should not be interpreted as the opinions of the Board, nor as official statements of LDS doctrine, belief or practice.
At a point in the appendix you state “The first consideration is that ancient depletion gilding was only able to remove copper and not silver, and when high concentrations of silver are present in a tumbaga alloy, the color no longer has “the appearance of gold” but tends to appear greenish or white (see figure 18).” Josiah Stowell, who saw a corner of the plate stack as it was being handed through the window, indicated that the plates had a greenish caste, which would indicate that the surface alloy (at least of that portion of the plates) had in the neighborhood of 25 percent silver and 30 percent gold.
Josh,
I noticed that in your article you indicated that the Encyclopedia Britannica reference must be wrong when it refers to tin being worked to as thin as 1/600th of an inch. On what basis do you make that argument? The article does not indicate that it is limiting itself to the boxed stackes of tin plate products. Tin was also sold and used (packaging, insulation, and even dentistry) in much thinner thicknesses than the boxed tin plate products. For example dentists in the 1800s commonly used tin in much thinner thicknesses than the commercial tin plate boxes to fill teeth. Extremely thin tin for dentistry was usually referred to as foil in the late 1700’s and 1800’s, but I don’t think there was a fine line in the manufacturing standards of the day that defined what was a plate and what was a foil. Of course one can make an assumption that the persons describing the plates were familiar with the boxed tin plate product specifications of thicknesses, but it is still just an assumption, even though perhaps reasonable. In any event, the reference from the Encyclopedia Britannica which you probably found in my Ziff book is not incorrect. My research determined it to be correct at the time based on the evidence and as far as I know is still evidenced to be correct. The testing that I did on the various thicknesses of the plates based on the rustling and flipping through technique did show that it was possible to go to .005 inches (or perhaps less). So I don’t think the “common tin” definition from the boxed tin products is absolutely definitive as to the minimum thickness of the plates.
hello jerry. first let me say that i very much appreciate your plate research!
as you know, 1/600th of an inch is .0016″ or less than 2 thousands of an inch. the encyclopedia entry states that “tin sheets” are “commonly” found to be this thin. there are three reasons i believe this is a misprint:
the first is because 2 thousandths of an inch is about three times thinner than the thinnest tin sheets specified in the modern ASTM standards (specifically, ASTM A624) which, of course, is based on modern manufacturing technology. tinplate this thin was unlikely to be even possible to manufacture in any kind of scale in the 19th century, and it was definitely not “commonly found”.
the second reason is that i have not found any evidence of any tin plate anywhere close to this thin in any 19th century industrial literature. the thinnest listing in burmingham wire gauge scale is twice as thick, and six times thicker than the thinnest tin plate standard of “one common”.
the third reason i believe this is a typo is because 1/60 (instead of 1/600) is .016″ which is the approximate thickness of common tin plate in the 19th century (about 27 BWG or 1XX) as indicated by the commercial literature.
–> all that being said, it is entirely possible that this encyclopedia entry is referring to what we would today call tin foil or aluminum foil rather than tin plates <–
and if this is the case, i would challenge the relevance of the data to the the 19th century "common tin" description of the plates. given the documentary record, it would be difficult to argue that the plates were made of a material similar to aluminum foil.
the only record which may support an argument for plates of similar thickness to aluminum foil could be an interpretation of the late recollection from emma smith, who said she recalled that the plates were similar in thickness to "thick paper" which is still over twice the thickness of 1/600th of an inch (see footnote 21) and significantly thicker than aluminum foil – and this is directly contradicted by a similarly late recollection by david whitmer who compared them to "common sheet[s] of tin used by tinsmiths" who generally did not work in foil, and when they did, it was relatively specialized. i suppose if the plates were like tin foil they would have been described more literally as "about the thickness of tin foil" rather than "common tin" or "tin plate" as appears in the documentary record.
so in summary, the entry is either a typo, or it's referring to what we would call "tin foil" which is not relevant to understanding the thickness of "common tin" in the 19th century as referenced in descriptions of the plates. of course, smart people can reasonably disagree on these sorts of things. 🙂
As I mentioned, your approach using the witnesses or secondary witnesses referring to “common tin” based on commercial grade classification of the time is not unreasonable, but some described it as a thickness less than common tin, so it would be also reasonable to assume the thickness was somewhat less than common tin. In my book I spent a little bit of time fleshing out window glass and common tin, but primarily deferred to my experimental results of the flexibility/rustling test. I recently looked at 19th century pasteboard (another description of Samuel Smith, a witness) and published in a podcast as follows:
“Although there are also various possible thicknesses for pasteboard, I obtained two different samples of 19th century common pasteboard, which was what playing cards and other common cardstock was made of.
The micrometer measured them at .011 or .012 inches thick.”
So people that use around .01 inches for the thickness are not necessarily wrong, but I think the general descriptions could put one down in the .008 inch range or so. As your paper says, there are various ranges of uncertainty about the description of the plates that end up providing a wide range of possible numbers of plates in the plate stack. Since I don’t really care about the apologetic and anti-Mormon polemics in my research, my attempt was just to take a scientific approach as you have done. In my book, in regards to the encyclopedia reference, I probably could have used the terms plate and foil instead of plate when referring to that reference. The point I was making is that tin could be made very thin, so an attempt to classify common tin thickness by assuming a minimum thickness was not going to be useful. Yours and others historical research is valuable to make an argument that the witness were referring to a commercial grade and classification, but who knows if their description of “common tin” was made with any technical knowledge of the tin plate classification system. That is an assumption. Some have said that anyone running a cooper’s shop would have known without understanding that iron was the metal used in those shops, occasionally copper and brass for specialized containers for things like gunpower, so that assumption is a very good one.
Anyway, I thought your paper was good, I did miss your FAIR presentation, hopefully can catch it on video.
Thank you for article. I’m currently writing a book on physical properties of the BOM plates per witness accounts (81 total): material; plate size and thickness; book thickness, weight, and binding; writing script, language; seal portion; and other characteristics. I handle the witnesses individually and then do collective reconstruction.
Some Initial Comments:
1. Plate size (Part 1, Step 1).
Witnesses described plates between 6″ square (S.F. Anderick, 1888) and 12″ square (Josiah Stowell, 1832). Most witnesses described a length between 7″ and 9″, and a width between 6″ and 7″–which are the same or very close to your figures.
2. Plate thickness (Part 1, Step 2).
It’s nice to know someone else figured out that of “common tin” refers to tinplate (See my comment on 3 Jan 2025 on Dan Vogel, “Joseph Smith Brings the Plates Home – Dan Vogel”, https://www.youtube.com/watch?v=mmX-H1GBivk). Thickness for tinplate ranged from about 0.3 to 0.9 mm (0.012 to 0.035 in.), making your minimum value of 0.005 in. in Step 2 too thin. Note that your footnote 21 to Barrett et al regards paper, not parchment.
3. Tumbaga (Part 1, Step 2).
You state: “These finding [gold and copper alloys in the Old and New World] provide historical context for the possibility that such an alloy was used in the creation of the plates.” To posit Tumbaga is problematic. There was no metallurgy in Mesoamerica (where most scholars locate the BOM) before 800 CE, plus tumbaga was used for jewelry and religious objects.
4. Void Space (Part 1, Step 4 / Appendix D).
Your fabricated discs (0.02 in) had a void ratio of 53%. That seems very high. I contacted Bill McKeever who has a plate replica and the void ratio of sheet metal plates (23 gauge, 0.027 in.) is about 8%. Because void ratio has a huge impact on the possible results, it requires more research/experimentation. Note that Jerry Grover in “Ziff, Magic Goggles …” (2015, p. 90) cited sheet fragments from Columbia between 800 to 1250 AD had a “measured variation” of 6.3% to 39%.
Your experiment used “pure copper rounds.” Half hard copper is likely a better match for ancient copper (Grover, “Ziff,” p. 89)
5. Writing properties (Part 2)
I’ll need more time to work through your data and then compare to:
* Neal Rappleye’s FAIR presentation, “Comparing the gold plates to known ancient artifact,” 15 November 2023, https://www.youtube.com/watch?v=eAgzJ8MYrcw
* Jerry Grover interview on Mormonism With The Murph, “Could the text of the Book of Mormon have fit on the gold plates”, 27 February 2025, https://www.youtube.com/watch?v=qPwOZ7ZSLBc.
6. Text Size (Part 2, Step 1). David Whitmer (1831) stated “some of them [“characters”] large and some small.”
ken, see comments below:
>2. Plate thickness (Part 1, Step 2).
Thickness for tinplate ranged from about 0.3 to 0.9 mm (0.012 to 0.035 in.), making your minimum value of 0.005 in. in Step 2 too thin.
yes, the 0.005 minimum is in reference to 19th century “parchment” – not tin. there are at least two references to paper-like materials used to compare to the thickness of the plates. for the analysis of tin, see table 3 and appendix E in the paper.
>3. Tumbaga (Part 1, Step 2).
You state: “These finding [gold and copper alloys in the Old and New World] provide historical context for the possibility that such an alloy was used in the creation of the plates.” To posit Tumbaga is problematic.
the plates, as described by the 19th century documentary record and based on the math, are unlikely to be made of 100% gold. the possible valid configurations of pure gold require very large voids between plates and very short stacks of plates, and the ones with tinplate thickness result in a number of plates that leave very little room for the complete text. these 100% gold configurations just don’t work very well.
so if not 100% pure gold, then this leaves us with an alloy, and i’m unaware of any other ancient alloy that can produce a finish with the “appearance of gold” aside from gold/copper alloy, commonly known as “tumbaga”.
>There was no metallurgy in Mesoamerica (where most scholars locate the BOM) before 800 CE, plus tumbaga was used for jewelry and religious objects.
i have no interest in speculating about specific book of mormon geography, but metallurgy (specifically tumbaga work) is found in the old world as far back as 900 BCE (Valdez 2005) and is found in the new world as far back as 2000 BCE (Hauptmann 2016) and recall that nephi was from the old world.
>4. Void Space (Part 1, Step 4 / Appendix D).
Your fabricated discs (0.02 in) had a void ratio of 53%. That seems very high. I contacted Bill McKeever who has a plate replica and the void ratio of sheet metal plates (23 gauge, 0.027 in.) is about 8%.
yes, 53% is very high compared to modern manufacturing flatness standards specified by the ASTME. bill mckeever used modern metal sheets manufactured to a level of precision which is not applicable to understanding ancient sheet metal.
>Note that Jerry Grover in “Ziff, Magic Goggles …” (2015, p. 90) cited sheet fragments from Columbia between 800 to 1250 AD had a “measured variation” of 6.3% to 39%.
as i read it, grover indicates that scott (Scott 2000) measured the thickness of ancient sheet metal and found a variation of 6-39% in its thickness, which does not tell you anything about its flatness. without a measurement of its flatness, variation in thickness will tell you nothing about the size of the void between stacked sheet metal. you would need to measure the variation in flatness, not thickness, and i would be very surprised if scott measured this, and i would even more surprised if the size of the metal he was measuring was anywhere near the size of a plate from the book of mormon. the larger the size of the sheet metal, the higher the void will be due to increased variability in the flatness.
>Your experiment used “pure copper rounds.” Half hard copper is likely a better match for ancient copper (Grover, “Ziff,” p. 89)
“half hard copper” is simply annealed copper, which is what was used in the experiment of the 165 copper rounds, and i’m not sure why you think annealed copper would be a better match for ancient copper? both ancient and modern copper are easily annealed as well as work hardened.
i would welcome additional experimental data on non-commercially manufactured flat plates and the void between stacks of them, but for now, the 165 copper disks is the only experimental data i’m aware of.
thank you for reading the paper, your feedback, and your interest in the subject. 🙂
Thanks for your quick response.
Void Space.
I think void space is an important factor which has not been explored. The reason I contacted Bill McKeever was to get the lower limit for the void space–nothing would be lower than modern produced plates which looks to be around 8%.
In your experiment, the copper rounds were “machine roll[ed], hammer[ed], anneal[ed], stress reliev[ed], and descal[ed].” Do you have any information about material properties (tensile strength, hardness), flatness of each sheet, weight applied during “manual compression.”
(Any expansion of Appendix D would be helpful.)
We’ll never know the exact void space, but as more people do research and experiments and analyzing the data, hopefully it will produce subsequent experiments that will nail down the likely range of the void space.
Ken, so far I would consider Josh’s experimental attempt at void space is better than my estimate in any event. The reality is that there is going to be a wide variety of surface deflection ratios and imperfections on ancient metallurgical plates. We just really have no statement from any of the plate stack descriptions as far as I know. Obviously to get an edge that one can flip through indicates a certain amount of separation. Other than that, I don’t know what else can be conjectured. The Narino plates are just one example. That issue is something that needs to be better studied and compiled as to the range in surface deflection in ancient flatwork metallurgy, I honestly I just don’t have time to do it.
Also, I noticed in the Ziff book that my editor for whatever reason neglected to put the citation on the surface measurements of the Narino plates, so at some point I will need to go back into my notes and sources I had 10 years ago and post an addendum as to the source. As I recollect it was a reference that was cited in the (Scott 2000) article but will have to go back and check. It is not the (Scott 2000) reference, as that was just for the general description of the Narino plates as indicated where the cite is placed. So if you are preparing a publication wanting to use that info, don’t cite Scott on that data.
The Ziff book was peer-reviewed, but as any who have published books can attest, alas some typos always make it through.
Good luck with your book, it looks like it will be a valuable compilation and contribution to the topic.
I don’t think your statement regarding information in my Ziff book here in your article that “variation in thickness will tell you nothing about the size of the void between stacked sheet metal” is incorrect to the extent it tells you “nothing”. It does provide some information that can be used to approximate minimum void space and maximum void space assuming a completely flat plate. For example, if one assumes that there there is a 39 percent variability at various points across the surface of the plate,, when such plates are stacked, you would have a 39 percent void space in the stack. I didn’t have any testing of the deflection of the ancient plates (which you refer to as flatness), so made an educated engineering guess that it could be 11 percent or so, so arrived at a 50 percent void space number. Your testing indicated that it ended up being somewhere in that range, so it seems like it was a reasonable test to help determine potential void space. So Ken Wick’s assertion that the Columbian disks is somehow a better test is not correct. Both tests are needed to help approximate a reasonable expected void space.
jerry,
i think it’s likely that we’ve misunderstood each other on this.
what i mean to say is that, for example, if i have 10 thin plates in front of me, and i take calipers and place them on the edge of each of them to measure their thickness and find that one is .05″ another is .07″ and another is .03″ and so on, then i can say that there is a variation between the thickness of the 10 plates (eg. the coefficient of variation between .03, .05 and .07 is about 32%) but this measurement will tell me nothing about their flatness, and so this data, by itself, can tell us nothing about the void between them when stacked.
my reading of scott referenced in your paper was that this is what he meant when he found a 6-39% in thickness – though it’s entirely possible that i’ve misunderstood what he’s measured.
in any case, your 50% void estimate aligns very well with the experiment data.
Jerry,
First, thanks for your efforts producing Ziff (and Josh for his article).
Like you, I realized that void space is a critical variable (see my original comment “Because void ratio has a huge impact on the possible results, it requires more research/experimentation.”; Grover, “Ziff,” p. 92, Scenario 1 comment “The primary variable that will affect the weight [of the plates] is the stack void.”).
In Ziff Scenarios 3 and 4 (pp. 92-3), you used 50% void space. This seems too high given the assumed plate thickness and the effects of weight flattening (0.005 in. for Scenarios 3 and 0.01 in. for Scenarios 4)? Meaning very thin plates, especially in Scenario 3, are pliable and the effects of weight flattening would decrease the void ratio the closer the plate was to the bottom of the plates. For example, in Ziff you commented:
* “In addition to the difference in thickness, there might be an overall deformation or slight warping unrelated to the thickness variation of an individual plate. … The warping or deflection would be minimized in a heavy stack of plates, so [void space] would not be expected to be great.” (p. 90)
* “As mentioned, there might be additional void space caused by larger areal scale deflection or warping of the Book of Mormon gold plates, but would probably not be extremely significant based on the weight flattening. The deflection in the bottom two-thirds of the stack would be expected to be minimal as the weight of the overlying plates would eliminate the effects of deflection or warping.” (p. 90)
Further, If a 5.9% to 39% void space is used based on thickness variation, how is this (or should this be) accounted for in the weight calculation (see Ziff, p. 90, and above 14 Aug 2025 comment “variability at various points across the surface of the plate”)?
Ken,
As you know I didn’t really get into the weight compression much in the Ziff book 10 years ago, that book was really an attempt to do some actual serious research on the metallurgy of the plates, primarily looking at determining the thickness of the plates and the methods to describe the gold gilding and black patina, as the weight issue of flattening on the bottom of the stack is going to be a bit complex and tbh kind of unknowable other than perhaps a range as we really don’t have the plates to examine (it would be nice) and the descriptions are varied. First, one can’t even scientifically look at the issue until the thickness of individual plates is determined (obviously the resistance to deflection by upward weight is partially determined by the thickness of a given plate) which I think my work (and others) gives a much more likely defined range (.008-.01 inches or so). The surface features or irregularities (as opposed to overall deflection) of the plates are important only to the extent that under weight-type compression, one would not expect the plates to compress much beyond the void space that results as a result of two uneven surfaces touching each other at high points of the irregular surface.
Probably the biggest complicating factor of weight compression is the whole issue of the sealed portion. It was described as “securely bound together” and “impossible to separate them” and as solid as wood and “united with solder” (Whitmer). As the sealed portion contains records compiled by Moroni, this lower section probably doesn’t exhibit any deflection of plates as it was created from smaller records by Moroni and has some sort of structural support from Moroni’s sealing process as individual plates are indicated that they couldn’t be moved. So then one has the issue of whether it is 1/3, or 1/2 or 2/3rd of the total stack based on the varied descriptions. That basically means that perhaps only the very bottom part of the top 1/3 would be subject to weight compression, which would not be that much. Of course certain critics generally like to assert that the sealed portion is 2/3rds, and then make the argument that the Book of Mormon can’t fit on the remaining plates. If one is wanting to argue that there is less void space in the stack (perhaps then reducing the number of possible plates) then one would argue that the sealed portion is 1/3rd of the stack. Kind of my recommendation to critics, who seemed to have more or less consolidated around the premise that there was some sort of sets of plates, would be to do research to try to recreate a set of plates that account for all of the descriptions, including the ability to thumb through them is some fashion. So far I haven’t see anyone produce much in the way of results in that area.
Anyway, in my opinion as far as void space, all anyone is going to arrive at is some sort of range of guestimate. Kind of like the engineering concept we engineers always joke about in that we sometimes measure it with a micrometer, mark it with a crayon, and cut it with an axe.
Hello Josh. Interesting paper. With regard to Appendix F and the “Lost 116 Pages” you indicate the D&C’s stating “more particular” as meaning a “more detailed” account in the small plates. I would then ask, how could Nephi be “more detailed” about Lehi’s writings, than Lehi was? I take the D&C to mean Nephi was more detailed about his own account of “these things” which would include abbreviation of Lehi’s Dream and blessings and expanding on those and a LOT of borrowing from Isaiah, but Don Bradley makes a compelling case in his book, “The Lost 116 Pages: Reconstructing the Book of Mormon’s Missing Stores,” for there being many more than 116 pages from the book of Lehi that were stolen. Would factoring in Bradley’s figures impact your findings? You ruled out anything over 116 pages when translation time is factored in. What if you ruled out everything less than 116 and nothing greater than 300 (Bradley p.95)? Just curious. Thanks for giving me some new things to think about.
Maggie, Josh will have to answer for himself. I will just note that we didn’t have Lehi’s writings. Even in 1 Nephi, Nephi abridged them. We might assume he did so on the other plates. However, the other plates were also more attuned to political themes, so perhaps they were not. In any case, the wild card is that we didn’t even have what Nephi wrote on the large plates, but only what Mormon thought was important. So, details could easily be lost.
maggie,
it’s certainly possible that “more particular” could mean something besides “more detailed” – though the examples of the term used in elsewhere in the book of mormon seem to indicate that it means something similar to “more expansive” though it’s up to interpretation.
i don’t know how many manuscript pages were lost – the traditional number is 116, however, as you noted, bradley has various estimates above this number – but it’s important to note that for the calculations in the paper the unit of ‘pages’ is too imprecise and so the unit of ‘characters’ is used instead.
so to account for various estimates of the length of the “116 pages”, the range of sizes that were included in the calculations start at *half* the size of the small plates of nephi (about 180,000 characters) all the way to *twice* the size of the small plates of nephi (about 720,000 characters.)
bradley suggests that a single page of the printers manuscript is equivalent to 1.15 pages of the modern printing of the book of mormon. assuming this, then the 116 pages is about 133 pages of the book of mormon, or just less than the modern printing of the small plates of nephi, which is a little over 142 pages in length. using this metric, this aligns with the idea that the lost pages were slightly shorter than the small plates of nephi, which also seems to be supported by the internal evidence in the scriptural text.
however if we assume the lost pages were closer to 300 manuscript pages in length, then in a modern printing they would be about 345 pages, which is over twice the length of the small plates of nephi.
the calculations that are included in my paper don’t quite go that far in accommodating an estimate that large, but they come close (about 284 pages instead of 345). see figure 7 in the paper for what this looks like.
but to answer your question, yes, there are many valid physical configurations of the plates that assume 345 pages in the lost pages. they just tend to have smaller text sizes, higher translation densities, thinner plates and more of them.
Just curious but would the method of inscriptions affect the size of characters? Such as a pointed scriber or a chisel with a mallet?
brian,
i have no particular expertise in the various ancient methods of metal engraving, but i have some experience in working with depletion gilded tumbaga. the outer layer of gold on a depletion gilded tumbaga item is relatively thin – just a number of microns in thickness, and making marks, such as characters, does not require the force of a mallet, but it is labor intensive and tedious.
as to how different methods of engraving may effect the size characters, i could only guess.
Josh, thanks for your article. While Moroni tells us that “we [likely Mormon and Moroni only] have written this record … in the characters which are called among us the reformed Egyptian,” he also tells us that the characters were “altered by us according to our manner of speech.” (Mormon 9:32) In other words, the characters that they used for writing were not the same as in the beginning of the record. While it is very possible that the authors of the small plates of Nephi used the Demotic script for their record, Mormon and Moroni many hundreds of years later certainly did not, as Moroni told us; they used an altered script that he called “reformed Egyptian.” So, there would have been at least two different scripts present on the plates: Demotic (possibly) and reformed Egyptian, whatever that looked like. Another reason that “reformed Egyptian” could not have been Demotic is because Moroni was clear to add that “the Lord knoweth the things which we have written and also that none other people knoweth our language.” (Moroni 9:34) Late Demotic script was still in use in Egypt at the time that Moroni scribed his final words.
loren,
i agree that the type of writing used in the small plates was probably different than the type of writing that mormon and moroni used hundreds of years later. i also agree that reformed egyptian is not demotic – but note that the analysis in the paper does not assume demotic was used on the plates. demotic, being a type of ancient egyptian written language, is simply a reference point for translation density in the analysis.
“YHWH” in the paleo-Hebrew 600 BC Ketef Hinnom inscription on silver amulets was about 0.3 inches long (7.62 mm) — https://library.biblicalarchaeology.org/article/words-unseen/ .
Your average estimate of 5mm square (3.9mm Table 13; 5.4 to 2.5mm Table 12) “with each character representing at least three English characters, similar to Egyptian Demotic” is absurd because Demotic does not always need more than one character (see cursive examples in Nibley, Since Cumorah, 2nd ed., 149). Hieratic and Demotic Egyptian can be used as “short-hand” (Champollion’s tachygraphie) in which one squiggle can represent an entire common phrase. “And it came to pass,” for example, is five English words which is written with one word in Hebrew (ויהי four Hebrew letters vocalized wayehi), but which could be presented in Demotic as one squiggle for ḫpr.f “it came to pass, it happened” (Chicago Demotic Dictionary H3 50-54), and which has a number of analogs in the “Caractors” Transcript. In Coptic it is written out as ⲁⲥϣⲱⲡⲉ ⲇⲉ “it came to pass.”
Naturally, names and important words could be spelled out in longer forms. The use of logograms would have saved space, which is the very rationale for using reformed Egyptian (Mormon 9:33) instead of alphabetic Hebrew.
robert,
ancient inscriptions on metal have been found to be as small as less than 1mm and as least as large as 9mm. tables 12 and 13 simply show examples of valid calculations which include 2.5mm and 5.4mm characters which are within the range found in extant ancient artifacts.
as for demotic, the paper is simply evaluating the average number of demotic characters and comparing them to the average number of english characters in an english translation of that demotic text. the fact that some demotic characters may represent more or less than an english word is not relevant to the calculation. see appendix B in the paper for more information on this.
This is incredible research! Thanks to Josh for putting this together
thanks ben. 🙂