The Voynich Ninja

(09-05-2026, 01:33 PM)Grove Wrote: You are not allowed to view links. Register or Login to view. “ The character 血 is matched with the EVA string aiinqot “
Hi Jorge, does this mean spaces are irrelevant in your theory?

I added that crib just yesterday, and I am still quite unsure about it.  

Spaces do seem to be relevant, since, on average, one Chinese character in the SBJ corresponds to one Voynichese token in the SPS, almost exactly.  That is the case not only for individual entries and for gaps between entries, but also for the histograms of entry length and parag lengths.

But spaces also seem to be often wrong, in both senses.  It must be often the case that a Chinese character corresponds to only part of a word in the transcription file, or to two or more words together.  How often does that happen? I do not know yet.

Besides the potential confusion with single-vowel syllables, which I mentioned in the previous post, there is also the fact that many concepts that are expressed by a single word in Western languages are expressed by two-character compounds in those candidate East Asian languages; and (as in any language) the meaning of the compound is usually related to the parts, but cannot be deduced from them.  Like 瘾疹 yǐn zhěn = "urticaria", literally "craving rash".  Nowadays, when Mandarin is transcribed phonetically as pinyin, the two syllables of such compounds are often stuck together without a space (yǐnzhěn), possibly to facilitate the parsing by Westerners.  So perhaps the hypothetical Dictator too sometimes read such compounds together as single words. Or perhaps the Author recognized that the two syllables were a compound, and joined them in his notes.

For that reason and other similar reasons, when I try to pair an SBJ recipe to an SPS parag, I ignore all spaces in the latter, and look only for substrings of EVA characters that seem to be correlated with the Chinese keywords characters.

In the case of  血 = "blood" (which, not surprisingly, is one of the most common characters in the SBJ), I am trying to see whether it can be another crib, the same way I found the others.  Namely, I took the SBJ entries that have I have already tentatively paired and that use that character, and looked for some common string that appeared around the corresponding positions in the SPS parags, modulo some "spelling" variations. 

On some (not all) of those parags, the string "aiinqok" seemed to appear at the right places, with some variations.  Currently I am assuming that the variations can be described by the RE pattern /aii?(in|r)q?o?[ktpf]/, that would match "aiirk", "aiiinqop", etc.   

The "?" and "(in|r)" in the pattern means that the string may start with aiin, aiiin, air, aiir; and that the "qo" is optional.

In a previous post, I explained why iin <-> ir is an acceptable variation; namely because they can easily look the same in moderately sloppy handwriting.  Said another way, the "ink distance" between the two is rather small. 

In another post I also explained why omitting the q or qo is also an acceptable variation.  Namely, because there is a vague hint that it is not phonetic but a symbol for "and", like Western "&", that the Author added because he felt it was necessary for comprehension.  (In a text like the SBJ, the phrase "apples, bananas, and cherries are good" would be written "林檎香蕉櫻桃美", literally "apple banana cherry good".  My hunch is that the Author found that quite confusing, and would write down "apple qobanana qocherry good" instead.)

Anyway, now I am trying to add 血 ≈ aiinqok (and variations) as a required keyword pair when matching entries that use that character. 

Maybe it will work,maybe not.  It may be that the Voynichese for 血 is not quite that string but some other string that may or may not overlap with it.  It may be that the entry-parag pairs were I got that string from were incorrect and the Voynichese for "blood" is nothing like it.  Maybe I will need to allow other variations, like o instead of a, or d replacing k (another pair of glyphs that are close in "ink distance"). Stay tuned.

All the best, --stolfi

Jorge, we are going in circles. You know full well that I don't think random chance explains the finding because I don't think the identification has happened by chance. You are at some pains to argue that the "trimming", editing, and many-to-many matching process that you have constructed is logically justified. Accepting all those arguments, which I don't, you posited a loose correlation between 主治|主 and [dklrs][ao]ii?i?[rn] because they are common and fairly evenly spread through the text. If I am reading your notation correctly, the Voynichese there corresponds to 120 different words and taken together they represent something around a high single digit percentage of tokens in the text. Yes, this is an astounding range of variation! The same can be said of 氣, which "only" has 24 associated words but includes the entire extremely common "ched" and "shed" "families" including their "ee" variations. You constructed this extremely permissive structure because of patterns you observed in order to explain those patterns, I do not think it is surprising that you are finding some matches.

By your own admission, this wide net has not been working as well as you hoped and needs some kind of refinement. It has not given way to any sort of hypothesis about Voynichese phonetics but it has generated a number of questions about why recurring phrases cannot be extended. This seems well in line with what I'd expect if these were spurious correlations because you'd overfitted the model to your hypotheses. I've laid out various other pieces of this in other parts of this thread.

As for Japanese, this is wrong, both as a description of the conversation here and of the language:

(09-05-2026, 08:58 AM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.I don't understand what you mean by "lacks finals". As @Jonas_Barnun explained, if the Dictator was Japanese, he probably read each Chinese character as a single syllable, with its "Chinese" reading.  So that "Japanese" language would be essentially another "dialect" of Chinese.

Jonas did not say this, and it is incorrect; onyomi are quite often two or three syllables. Finals are the Chinese rhymes/rimes and one of the foundational idea behind all attempts at creating descriptions of the phonology and alphabets. There are equivalents, and the term is used to describe them, in most languages throughout the MSEA language area. The complexity of these finals, and the ways they get Romanized, significantly contribute to the lower entropy of texts Romanized to describe them. Japanese does not have these complex finals, and when the onyomi were borrowed, the checked tones were reanalyzed as multiple syllables. It results in a very different sort of phonology, and one that fits the Latin alphabet remarkably well and was natively written with just 48 characters made for expressing foreign sounds quickly.

Otherwise, I'm not sure I have the thread of our disagreement here. I'm not contending that you need to have an automatic process for decryption, so I'm not sure why that has come up. Asking for any kind of description of how Voynichese works and bridges with a specific text (not the idea of a family of texts) falls well short of that demand. If the COT and this putitive match with the Shennong Bencaojing are independent, that would explain why you are maintaining Japanese as a possibility, but in that case, why not simply drop the contention that Voynichese is monosyllabic? You equivocate enough on the spaces I'm not sure if you're maintaining them or not. Like, you plainly say you are, but then seem to defend the idea that you can ignore them after all? If you're not practically respecting them in your algorithm---and this is another thing, you're the one automating the search with an algorithm, which you say is a misguided conception of the problem?---are you not disregarding spaces in practice?

The reason I brought up the rhetorical perspective is that I'm struggling to fit all these positions into a coherent position. I have problems with many of the pieces, but I am struggling to articulate what you actually think the case for this being a "Chinese" text is. It's not great that every time we go around on this we get bogged down in questions like, "Was the Gregorian calendar in use in 1420?" and "is Japanese monosyllabic?" and "is it still vaginal discharge if you forget to mention it's women who are experiencing the discharge?" But also, I am not clear which arguments you're maintaining and if you think they actually matter. Rhetorically, why not embrace the irrelevance of the COT if you think the SBJ identification doesn't hinge on it? What rhetorical purpose did all that energy you spent defending the Chinese interpretation serve once you pointed out you didn't need to maintain it to defend your proposal of a Chinese origin? Which positions that you've made matter and which ones should we be weighing here?

(09-05-2026, 11:38 PM)rikforto Wrote: You are not allowed to view links. Register or Login to view.Jorge, we are going in circles. You know full well that I don't think random chance explains the finding because I don't think the identification has happened by chance.

You are right, we are going in circles.  It seems that you will never accept anything other than a perfect solution -- without ambiguities, assumed errors, variant spellings, etc. 

Well, I believe you will be disappointed.  Even if the SPS is not the SBJ and the COT is false, the solution will have to look just like this.

So I can only say that you should stop reading this thread now, and come back only when I am done -- or I have given up.

Quote:[Your approach] has not given way to any sort of hypothesis about Voynichese phonetics

Yes, you must admit that it is a situation that no one expected.  I have the complete translation of the SPS, including the four missing pages, and know (as well as possible) who is the author and when and where it was written.  But I still don't know the language and the spelling system.

Quote:As for Japanese, this is wrong. [...] Finals are the Chinese rhymes/rimes and one of the foundational idea behind all attempts at creating descriptions of the phonology and alphabets. There are equivalents, and the term is used to describe them, in most languages throughout the MSEA language area. The complexity of these finals, and the ways they get Romanized, significantly contribute to the lower entropy of texts Romanized to describe them. Japanese does not have these complex finals, and when the onyomi were borrowed, the checked tones were reanalyzed as multiple syllables.

Sorry, I still don't understand your objection.  While some on-yomi readings like 学="gaku" now have two or more syllables, most have only one.  (And maybe in 1400 a Japanese scholar may have read 学 as "gak" not "gaku".  Who knows.)  

Anyway, this point is not important. @jonas_barnun observation just added Japanese on-yomi to the list of 50+ candidate languages.

Quote:If the COT and this putative match with the Shennong Bencaojing are independent, that would explain why you are maintaining Japanese as a possibility, but in that case, why not simply drop the contention that Voynichese is monosyllabic?

Because the structure of the Voynichese words strongly points that way.  Independently of the COT or the SST.  And, again, Japanese on-yomi is mostly monosyllabic.

Quote: you're the one automating the search with an algorithm, which you say is a misguided conception of the problem?

There will not be a *decription* algorithm, that takes an arbitrary Voynichese text and outputs a plaintext in some known language.  

There will not even be a *validation* algorithm that verifies whether a given plaintext is a valid "decryption" of a given Voynichese text.

What I have is a *matching* algorithm that tries to identify which of 243 parags of the SPS is the Voynichese rendering or a given entry of the SBJ, given the Chinese text of the latter. It does not give a categorical answer, but only a measure of the badness of fit, based on the discrepancies between the spacings of certain keywords in both texts.  

When I am lucky, it finds one match that is much better than all the others.  

Other times it does not find any decent match.  Considering that I can search only 243 of the estimated 365 parags (~67%), this should happen at least 33% of the time.  Depending on the keywords that I specified, it may not find any match at all.

Sometimes it finds two or three parags that match the entry reasonably well.  This typically happens with entries of middling size (which therefore may have half a dozen candidate parags of about the right size) and that have only one or two occurrences of the current set of keywords.  Since the entries have a very rigid structure, it does happen that two entries have the same number of diseases, and thus their 主治 keywords end up in about the same positions.  I am counting those cases as failures, until I can find more cribs that will let me refine the match.

Right now, of 52 entries I tried to match, 22 failed (42%), of which maybe half a dozen are multiple matches. So obviously I didn't give myself enough slack...

Quote:Rhetorically, why not embrace the irrelevance of the COT if you think the SBJ identification doesn't hinge on it? What rhetorical purpose did all that energy you spent defending the Chinese interpretation serve once you pointed out you didn't need to maintain it to defend your proposal of a Chinese origin? Which positions that you've made matter and which ones should we be weighing here?

I became convinced of the COT around 2000, because of the statistics of the text -- including the distribution of lengths of word types and their structure.  Since that time, its central claim was that Voynichese was an East Asian monosyllabic language in a phonetic encoding, not encrypted.  The story about dictation to a traveler was only meant to show that the central claim was plausible.  Besides the statistics, there were other bits of clues supporting the COT, such as the Zodiac with 24 equal parts and the big red glyphs on f1r.  That evidence was enough for me, but evidently not for anyone else.

I don't recall when I learned about the Shennong Bencaojing and noticed the coincidence of its reported number of entries (365) and the estimated number of parags in the Starred Parags section (300-400). It was many years ago.  At that time, SPS=SBJ became just a side guess attached to the COT.  But back then I had lost interest in the VMS because the COT meant that I lacked the knowledge needed to go any further.

It was only very recently that I got a digital copy of the SBJ and (thanks to Unicode, and Unicode support in python) could verify that the histogram of recipe sizes matched quite well the histogram of SPS parag sizes, after adjusting for the lost and mangled parags.  

And it was only even more recently that I figured out the first "crib"  主治 = daiin by comparing the longest SBJ recipe with the longest SPS parag.  

As I explained, strictly speaking, logically the SPS=SBJ claim is independent of the COT.  But obviously is a huge rock of evidence in its favor.

Now let me say this: of all the theories I have seen since I "discovered" the VMS in 1998, the COT is the only theory that [a] is compatible with everything we see in the book and everything we know of its history, [b] is historically, psychologically, and pragmatically plausible, and [c] satisfactorily explains why someone decided to create a manuscript like this, and spent so much time on it. 

The various hoax/gibberish theories, for instance, fail [a] and [b] in several ways, and fail completely at [c]: they cannot explain why the Author would have decided to create a fake book like this.  They are like claiming that a banknote forger would spend 6 months carving and etching a crude master stamp to produce some fake Monopoly money.

All the best, --stolfi

I was checking up on something in a response of a very different character, and an oddity lead me to wonder about the transmission history of the Shennong Bencao Jing. That lead me to discover something interesting---this text didn't exist in 1420!

It's not a small detail that this was a lost text in 1420, right? Your little story about a scribe asking "what's your most important materia medica" and getting directed to the Shennong Bencao Jing runs into the not insignificant problem that there were no copies of it at the time. You are not allowed to view links. Register or Login to view. says that the first reconstruction was by a man named Miao Xiyong, and You are not allowed to view links. Register or Login to view.puts that publication in 1625. Admittedly it's got a weird status as a lost text because the contents were still in circulation, and that's going to turn out to be the detail that all this turns on, but at face your analysis could be finding the chunks that made it into the critical editions you're looking at. But even before seeing if that holds up to deeper scrutiny, this seems really important for analyzing your claims and adds to the various ways that you have not identified, let alone translated, the VMS.

The actual work, in the sense of a materia medica attributed to Shennong, vanishes sometime in the Tang dynasty, which is far enough back that positing a millennium of undocumented transmission is beyond speculative. The reason it fell out of circulation was because it had been expanded and significantly annotated in Tao You are not allowed to view links. Register or Login to view., and that transmission formed the main line of evidence for reconstructing it---again, well after the VMS carbon dating. Descriptions lead me to think that chunks of the SBJ were pretty recognizable, but actually lining up the received texts makes me think that's not the case, at least for the Rooster entry. The SBJ is on top and the text available in 1420 on the bottom:

Quote:丹雄雞
丹雄鸡

味甘　微溫。　　　　　　主治女子崩中漏下，赤白沃，補虛，溫中，止血　　　　　　通神，殺毒，闢不祥。頭，主殺鬼。
味甘，微温、微寒，无毒。主治女人崩中漏下。赤白沃，补虚，温中，止血。不伤之疮，通神，杀毒，辟不祥。头：主杀鬼，东门上者弥良。白雄鸡肉：味酸，微温，主下气，治狂邪，安五脏，伤中，消渴。乌雄鸡肉：微温。主补中，止痛。胆：微寒，主治目不明，肌疮。

肪，主治耳聾。　腸，　　主治遺溺　　　　　　　　　　　　　　　　　　　　　　　　　肶胵裹黃皮，　　　主治洩利。　　　　　　　　　　　　　屎白，　　　主治消渴，傷寒　寒熱。
肪：主治耳聋。鸡肠：平，主治遗尿，小便数不禁。肝及左翅毛：主起阴。冠血：主治乳难。　　里黄皮：微寒，主治泄痢，小便利，遗溺，除热，止烦。屎白：微寒。主　消渴，伤寒，寒热，破石淋及转筋，利小便，止遗溺，灭瘢痕。

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　翮羽，主下血閉。　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　雞子，　除熱火瘡，癇痓，可作虎魄神物。
黑雌鸡：主治风寒湿痹，五缓六急，安胎。其血：无毒，平。主治中恶腹痛，及　折骨痛，乳难。翮羽：主下血闭。黄雌鸡：味酸、甘，平。主治伤中，消渴，小便数不禁，肠　泄痢，补益五脏，续绝伤，治虚劳，益气力。肋骨：主治小儿羸瘦，食不生肌。鸡子：主除热火疮，治痫，可作虎魄神物。卵白：微寒，治目热赤痛，除心下伏热，止烦满，咳逆，小儿下泄，妇人产难，胞衣不出。醯渍之一宿，治黄胆，破大烦热。卵中白皮：主久咳结气，得麻黄、紫菀和服之立已。鸡白蠹：能肥脂。生朝鲜平泽。


(No aligning text)
鸡此例又甚多，云鸡子作虎魄者，用欲 卵黄白，混杂煮作之，亦极相似，惟不拾芥尔。

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　雞白蠹，　　　　　　　　　　　　　肥豬。生平澤。
又煮白合银，口含须臾，色如金。鸡子不可合葫、蒜及李子食之。乌鸡肉，不可合犬肝、肾食之。小儿食鸡肉，好生蛔虫。又鸡不可合芥叶蒸食之。朝鲜乃在玄兔乐浪，不应总是鸡所出。今云　白蠹，不知是何物，恐此别一种尔。

Note that the SBJ is traditional characters and the Tao's text are simplified, but I believe they are faithfully aligned. You probably need to copy this into a text editor without word-wrap to really see what I'm getting at here, but I did check and that does align them for me at least; apologies, I know that is a little cumbersome.

From what I can see in English descriptions of the history of the reconstruction, there were other, fragmentary lines of evidence collected later to help separate the two, and more recent discoveries have shed light on the transmission since. But unless you have evidence I am not finding, it seems like this was a dead tradition before 1000, and the SBJ you are using wasn't created until a much later date. At the very least, it seems completely out of the question to claim this was the go-to materia medica in the Sinosphere in the 1400s, something I had been taking at face value

(11-05-2026, 06:03 PM)rikforto Wrote: You are not allowed to view links. Register or Login to view.It's not a small detail that this was a lost text in 1420, right? Your little story about a scribe asking "what's your most important materia medica" and getting directed to the Shennong Bencao Jing runs into the not insignificant problem that there were no copies of it at the time. [...] The actual work, in the sense of a materia medica attributed to Shennong, vanishes sometime in the Tang dynasty, [...] Descriptions lead me to think that chunks of the SBJ were pretty recognizable[...] it seems completely out of the question to claim this was the go-to materia medica in the Sinosphere in the 1400s, something I had been taking at face value.

No, the text of the Shennong Bencaojing was never "lost".  By ~1400 CE it was no longer available as a book by itself.  But it has always been revered as the foundation of traditional Chinese medicine, and it was included in its entirety in popular medical texts of the time.  Much like the Pentateuch has long ceased to exist as a separate book, but has always been available as part of Jewish and Christian bibles.  

For example, in the Zhenglei Bencao (1108 CE) the text from the original 365-recipe Shennong Bencao was clearly distinguished by color (white on black instead of black on white) from the much more voluminous commentaries and additions.  The modern "reconstruction" consisted basically in extracting the parts of those texts that are marked off as "Shennong Bencao" and fixing the errors, omissions, and interpolations that had creeped into them.

For all I know, if the hypothetical Marco Polo 3.0 had asked a doctor in China for the most important text of Chinese medicine that he could copy, the guy would have taken out the Zhenglei Bencao or some similar alfarrábio(*) and pointed to the reverse-colored parts.

Moreover, the Shennong Bencao had spread to other countries in the Chinese area of cultural influence, like Vietnam, Korea, Japan, and Tibet (not just within the Chinese "empire") at least 1000 years before the VMS was created.  In Korea, the Shennong Bencao was mandatory textbook for all medical students.  In those countries the text was still printed in Chinese characters, but read either in some pseudo-Chinese, or in words of the local language, or somewhere in between the two methods.

All the best, --stolfi

(*) Portuguese has a couple hundred or so words of Arabic origin, many of them distinguished by the initial al-, ar-, or az-.  "Alfarrábio" is my favorite. It is a somewhat disparaging term for a ponderous old book that only scholars would read. The word comes from Al Farabi, a famous philosopher who lived in Baghdad around 900 CE, under the Abbasid Caliphate.  He seems to have been a quite decent guy (called by some "the second Aristotle"), who did not deserve that "honor"...

(11-05-2026, 08:21 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.For example, in the Zhenglei Bencao (1108 CE) the text from the original 365-recipe Shennong Bencao was clearly distinguished by color (white on black instead of black on white) from the much more voluminous commentaries and additions.  The modern "reconstruction" consisted basically in extracting the parts of those texts that are marked off as "Shennong Bencao" and fixing the errors, omissions, and interpolations that had creeped into them.

This should strike you as strange, I think. Qing scholars, for all their foibles, were not idiots, and a multi-generational reconstruction effort arguing about how best to recreate the text would not be necessary if they basically just had to transcribe some red letters they all had in their books.

Indeed, if you read the source I linked, it had a very interesting detail:

Quote:[Miu had a couple of friends sharing the interests in Chinese materia medica. The most distinguished among them is Lu Fu, a leading medical figure in Hangzhou. Lu spent over a decade separating and salvaging quotations derived from the original Classic scattered across multiple sources and constructing them into a single book. Although the quality of this edition is not highly regarded, he is remembered as the first medical scholar who attempted to restore the original texts of the Classic in late Imperial China.]

It would seem that someone trying to bring all the red text into a coherent original using a network of people collecting relevant editions did not arrive at a well-regarded critical edition. That doesn't imply they just had to copy some red text and be done with it, but rather that there were substantial gaps in the transmission.

Confirming this is, admittedly, a bit of a bear and I still have some questions about the exact state of the corpus in 1420. But this machine translated description of You are not allowed to view links. Register or Login to view. from the Ming era gets at the flavor of what my search found:

Quote:'Bencao Chengya Banjie' is a ten-volume work on Ming-dynasty Lu Zhiyi's compilation of Chinese materia medica. Lu Zhiyi, a notable scholar and physician, aimed to restore and elaborate on the 'Shennong Benjing' (神农本经), which lists 365 medicinal substances, corresponding to the number of days in a year. According to Lu, these 365 entries were not to be altered or reduced. However, one-third of the substances listed in the 'Shennong Benjing' were no longer available, so Lu selected 222 entries from the original text and added 143 more from subsequent compilations, ranging from Tao Hongjing's 'Bielu' (别录) to Li Shizhen's 'Bencao Gangmu' (本草纲目), to maintain the total of 365 substances.

A quick search of that text did not reveal a single mention of a chicken (the word "rooster" contains "chicken", and You are not allowed to view links. Register or Login to view. just straight up calls the relevant entry "chicken"). You're welcome to make a more thorough canvas of the text, but at the very least, it demonstrates the Ming understanding of the SBJ was quite different during the Ming era. Relatedly, if you can find Lu Fu's compilation, I would like to see that as well, but it's poorly regarded enough that I had trouble digging up a copy I could access.

The English language claims that reconstructions relied on red text seem to be true to a point, but based on much older texts that You are not allowed to view links. Register or Login to view.. That source is also of interest because it shows the transmission of the differentiated red text breaking down in the Tang era, though not entirely. There is no era where these rubrications were a universal priority, and that's unsurprising given that red ink is much more expensive than black, which was already quite expensive. It does seem to me that the red text convention was still around in 1420, but it seems like it had either degraded substantially through transmission or changed use (e.g., only marked out the titles of sections with no respect for ancient sources). I am inclined to trust that our Ming sources didn't just fail to copy some red ink, and that our Qing sources didn't undertake delicate comparisons because everyone essentially had versions of the SBJ just lying around and could read them off themselves. 

We can check the received text of the Zhenglei Bencao, in this case its own You are not allowed to view links. Register or Login to view., and see if our Ming scholars were as incompetent as implied by the availability of this text

Quote:丹雄鸡　味甘微温微寒无毒主　女人崩中漏下　赤白沃　补虚　温中　止血久伤乏疮通神　杀毒　辟不祥　头　主杀鬼　东门上者尤良白雄鸡肉味酸㣲温主下气疗狂邪安五藏伤中消渴乌雄鸡肉㣲温主补中止痛　胆㣲寒主疗目不明肌疮心主五邪血主踒折骨痛及痿痹肪　主　耳聋　肠　主　遗溺小便数不禁　肝及左翅毛主起隂冠血主乳难　肘胵裹黄皮㣲寒主　泄利小便利遗溺除热止烦尿白微寒主　消渴　伤寒寒热破石淋及转筋利小便止遗溺灭瘢㾗黒雌鸡主风寒湿痹五缓六急安胎　血无毒主中恶腹痛及踒折骨痛乳难翮羽　主下血闲　黄雌鸡味酸甘平主伤中消渇小便数不禁肠澼泄利补益五藏续絶伤疗劳益气肋骨主小儿羸瘦食不生肌鸡子　主除热火疮　癎痉　可作虎魄神物臣禹锡等谨按药对云鸡子平卵白㣲寒　　　　疗目热赤痛除心下伏热止烦满咳逆小儿下泄妇人産难胞衣不出□渍之一宿疗黄疸破大烦热　卵中白皮主久咳结气得麻黄紫菀和服之立也鸡白肥脂生朝鲜平泽

丹雄雞　味甘微溫。　　　主治女子崩中漏下，赤白沃，補虛，溫中，止血　　　　通神，殺毒，闢不祥。頭，主殺鬼。　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　肪，主治耳聾。腸，主治遺溺。　　　　　　　　　　　　　　　　　　　肶胵裹黃皮，　主治洩利。　　　　　　　　屎白，　主治消渴，傷寒寒熱。　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　翮羽，主下血閉。　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　雞子，　除熱火瘡，癇痓，可作虎魄神物。　　　　　　　　　　　雞白　蠹，肥豬。　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　生　　平澤

You will once again have to copy this to a text document with word wrap turned off and deal with the competing script standards, but the effect is once again quite striking if you do. (Also, if you check the underlying document, it was not text in red, but rather larger text---but this is no great matter, that style is well-documented.) I would say a few interpolations had crept in, yes. The straightforward conclusion I draw from this comparision is that attribution to Shennong was important, but the text itself was quite buried.

It is a testament to Sun Shinyan's scholarship that the Turfan findings were mostly confirmatory, but from what I can tell the version of the SBJ you're relying on was not accessible in a single compilation that was in circulation during the Ming Era, and interest in it doesn't pick up until after the VMS anyway. For your purposes, it seems like this text was lost.

Moving Northeast from China puts you more in my wheelhouse:

(11-05-2026, 08:21 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.In Korea, the Shennong Bencao was mandatory textbook for all medical students.  In those countries the text was still printed in Chinese characters, but read either in some pseudo-Chinese, or in words of the local language, or somewhere in between the two methods.

I was able to trace the claim that it was adopted in Silla to You are not allowed to view links. Register or Login to view., but I cannot find the underlying source. (The citation appears to be entirely translated and I cannot recover what the original was supposed to be.) I pretty quickly found in the You are not allowed to view links. Register or Login to view. that there is a lively debate which materia medica was adopted by Silla, but it wasn't the SBJ either way. Taken with the fact that the Tang government was involved in the Silla state at that point and officially backing other materia medica, I think the English source is in error, though whether or not it originates it I cannot say. At any rate, Silla falls in 985 and there are pretty radical shifts on the peninsula. I cannot find any Korean language sources suggesting it was still a prime text in the Goryeo period, and a cursory look makes me think You are not allowed to view links. Register or Login to view.. I see Japanese fragments (to be clear, they are Chinese language, but from fragments in Japan) used in the 1800s to reconstruct the texts, but this appears to be the same kind of multi-source critical work that Sun was doing. The picture I have is that this tradition was largely buried in Korea and Japan by the second millennium.

Can you show evidence that these transparent passages of the SBJ were still circulating and reliable enough to have made it into the VMS in line with the text you are using?

(12-05-2026, 05:29 PM)rikforto Wrote: You are not allowed to view links. Register or Login to view.We can check the received text of the Zhenglei Bencao, in this case its own You are not allowed to view links. Register or Login to view., and see if our Ming scholars were as incompetent as implied by the availability of this text

Please clarify this quote.  (Keep in mind that I don't know Chinese; I know only a few dozen hanzi and efw bits of the grammar -- mostly, what it doesn't have.)

I recognize the bottom five lines as an exact copy of the text that I am using, except for the punctuation.  It seems that parts of it are embedded in the top seven lines.  But, in those lines, how can I tell what is the quoted SBJ and what is Zhenglei's interpolations, without the font colors? I can see that many hanzi match verbatim, but it would help if the allegedly SBJ parts were marked off, as in the printed book...

EDIT: After mapping both texts to simplified characters, and comparing the hanzi by hand, I found that practically every character of my version (bottom 5 lines) was present in the ZLB version (assumed to be the top 7 lines), in the same order, intercalated with lots of other material.  After deleting the latter, I asked Google AI to compare the two lines.  It pointed out what it claimed to be several errors in the ZLB version, some in the book, some allegedly introduced when the book was digitized.  In all cases except the last it said that my version (which I suppose it did not know was "mine") was the correct one, more faithful to the original SBJ.  The errors it noted where:

1. Your ZLB: 女人 (woman, modern), my SBJ: 女子(woman, archaic) 
   
2. Your ZLB: 肘胵 (elbow-gizzard), my SBJ: 肶胵 (gizzard)
   
3. Your ZLB: 尿白 (white urine), my SBJ: 屎白 (white feces)
   
4. Your ZLB: 血闲 (blood leisure), my SBJ: 血闭 (blood blockage)
   
5. Your ZLB: 癎痉 (epilepsy, modern), my SBJ: 痫痓 (epilepsy, archaic)
   

It then says that both versions of the final sub-entry are wrong. The correct text (it says "according to Tao Hongjing’s compilation"), would be 
鸡白蠹：能肥脂 which it translates as "chicken[-coop] white grub: makes one's fat".  However, the same Google AI, a couple of months ago, when I asked about that sub-entry, gave a different hanzi version that it translated as "fattens pigs" and explained that it was probably some farmer who added that marginal note and it got copied into later editions.  Sigh.

There are also several other differences that,although don't change the meaning, can make a difference to my matching attempts.  It noted that your ZLB version (like the original SBJ, it claims), used only 主 to introduce a list of uses, whereas my version uses sometimes 主 ("mainly [for]") sometimes 主治 ("main uses") .  And indeed one problem I was having is that both 主 and 主治  seemed to map to "daiin" in the VMS.  But if the VMS was copied from a more faithful source, like the ZLB, then it all makes sense:  主 = daiin, always.

And it also noted that in your ZLB version there is a 主 after 鸡子 "eggs", whereas in my SBJ version there is none.  But the VMS does have a daiin there!  So now that makes sense too.

Thanks for the ZLB quote, and all the best, --stolfi

In fairness, I think you can blame my English here---I ought to have been more explicit. The top line is the part of the Zhenglei Bencao that the Qing editor attributes to Shennong. I already cut all the text that was set off as commentary; the original entry was quite larger still. In turn, yes, the bottom line is the reconstruction of the Shennon Bencaojing you're familiar with. As you can see, the discrepancy is substantial!

I also forgot to mention that the version of Tao You are not allowed to view links. Register or Login to view. I cited above does not make any distinction, and frustratingly does not have an underlying text there for me to check. A quick search did turn up some of the very old hand-penned texts that had the red ink, but all the older printed editions seem to lack any of the distinctions for marking out Shennon's text. You begin to see the problems a Ming scholar tackling this problem would have had as well!

(12-05-2026, 05:29 PM)rikforto Wrote: You are not allowed to view links. Register or Login to view....

I just edited my previous reply.  Please check it again.

All the best, --stolfi

(13-05-2026, 12:33 AM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.[Google AI] then says that both versions of the final sub-entry are wrong. The correct text (it says "according to Tao Hongjing’s compilation"), would be 
鸡白蠹：能肥脂 which it translates as "chicken[-coop] white grub: makes one's fat".  However, the same Google AI, a couple of months ago, when I asked about that sub-entry, gave a different hanzi version that it translated as "fattens pigs" and explained that it was probably some farmer who added that marginal note and it got copied into later editions.  Sigh.

I pressed Google AI again about that sub-entry.  It admitted that a scholar or doctor by ~1400 CE would have got that "rooster" entry from the Zhenghe Bencao, which (it claimed) read 鸡白蠹：肥脂 .  The last two characters, it said, meant "makes one's fat fatter" rather than "fattens pigs", a reading that was due to someone mistaking the hanzi of "fat" for that of "pig".

Fine. But then it complained that I my indentation of that entry had implied that the provenance field［生］平泽 applied to the whole entry, and said that it instead applied to those 鸡白蠹 only, and that the provenance was actually from Joseon, and thus that the 鸡白蠹 were actually grubs of a wild beetle imported from Korea.  

Then asked why were the grubs described as 鸡 = "chicken [coop]" if they were wild insects.  It explained that the "coop" part was a common misconception, and the 鸡 actually meant that the grubs just resembled chickens.

Then I asked why that sub-entry was included in the red rooster recipe at all, if it was about grubs that grew in the wild and not even in chicken coops.  It complimented me for my shrewdness and explained that it was because the Han Dynasty philosophers believed that, since those grubs resembled chickens, they were spontaneously generated in the wild by influence of the Qi from the chickens. That is why the Shenong Bencao authors felt that they belonged in the "rooster" entry.

Then I thought of asking whether it had any source for that theory that chicken Qi generated grubs in the Korean wilderness.

It admitted that it had no sources and had just deduced the explanation from the Han Dynasty view of biology etc.

Then I decided that the best solution was to get a beer and watch some soothing You are not allowed to view links. Register or Login to view. on Youtube.

All the best, --stolfi