QUERY> On the term 阿傍

Eric Greene's picture

Here is an old-timey H-Buddhism query, of the kind that does not seem to get asked so much anymore ever since replying to messages became more complicated that just replying to an email (I digress).

Does anyone have a decent interpretation of the Chinese Buddhist term abang 阿傍 (var. 阿旁 and some others too)? Anyone who has spent time reading Chinese Buddhist accounts of the netherworld will be familiar with this word I suspect. What it means is not in doubt: the guardian/jailors of hell/purgatory, usually depicted with bull heads (and the term 牛頭阿傍 is quite common too). But the question is from where the term comes and what it  means literally.

Nearly a century ago (!) volume 1 of the Hobogirin (p.8-9) noted that the word “has the appearance of a transcription [from an Indic term] but the original  is unknown.” I have looked in many (though not necessarily all) other modern dictionaries and reference works in various languages and none gets more specific than this. Nor have I yet encountered any modern scholar, when discussing/translating a text with this word, give any kind of explanation beyond noting what it points to.

Sparing you the full details of the notes I’ve accumulated about this word over the years, I’ll just also say, to begin, that despite the ubiquity of the word in Chinese accounts of Buddhist hell, and despite the clear presence of the referent (bull-headed hell guardians) in Indian Buddhist sources, the term 阿傍 itself (though again, not what the term points to) seems to be somewhere between rare and exceedingly rare in Chinese Buddhist literature unquestionably translated directly from Indic sources (The Hobogirin entry notes one instance in the Mahīśāsaka-vinaya, Wu fen lü 五分律, T.1421:22.184c18–22; that turns out to be perhaps the only indubitable case).

So maybe it’s not a transcription at all? Or…?

Eric Greene

Department of Religious Studies

Yale University

451 College St. 

New Haven CT, 06511



  Hello Eric!
  As a Sanskritist, I feel indeed that it is not a transcription of an Indic term. My gut feeling is that this could be one of two things:
  1) A compound of the kinship/honorific ā- 阿 followed by the verb 'to assist', yielding "assistants" or "helpers." I have the impression that I have seen that nominalizing function of ā- 阿 in Buddhist Chinese, but right now I can't recall where exactly, hopefully I will do soon.
  2) A nominalized verbal phrase with basically the same meaning. ēzhù 阿助 occurs in the Xin Tang Shu (奸臣傳下,崔昭緯) with a verbal force (每它宰相建議,或詔令有不便於己,必使鋋密告行瑜,使上書訾訐,己則陰阿助之). What we have here is ē 阿, the verb "to follow, obey." So perhaps "ābàng" 阿傍 is actually ē-bàng "[those who] follow and assist [the king of hell, the infernal bureaucracy, etc.]"?
  I love this kind of lexicographic problem, although I am not sure I was very helpful in this case!

Dear Eric and Diego,

I have to admit this is something bugging me for a while, because the exact same term 牛頭阿旁 also shows up in Tibetan materials as "a ba glang mgo" ("bull-headed Aba") in the same context. In the Tibetan canon, it shows up only in the mNgon par rtogs pa mu tig gi phreng ba (Q 5022, not in Derge) several times, e.g.,

lhor a ba glang mgo / dud kha lcags kyu dang zhags pa thogs pa /

(Q 5022 might or might not be a genuine translation of an Indic text; I have no way to tell)

I have not digged deeper, but the earliest Tibetan writer who mentioned this name seems to be kLong chen pa. It also shows up in the writings of Dalai Lama V and others.

So far I have not been able to locate any reference to a narakapāla described as either aśvaśīrṣa or gośīrṣa (gavaśīrṣa/gavayaśīrṣa/gomukha) in any Sanskrit text. (Maybe both 牛頭馬面 and 牛頭阿旁 were invented outside India?)

Best wishes







   I would add that the Tibetan a ba could be consistent with a transcription from Chinese. The Chinese final was typically not rendered in Tibetan and Khotanese renderings, and this is probably the outcome of the manifestation of that final as nasalization of the preceding vowel rather than as a consonantal coda in the northwestern dialect of Middle Chinese that would have been current in the Western Regions, e.g. 傍=Northwestern *bã rather than *baŋ. A few examples from the Pelliot Tibetain texts considered by Walter ("A Note on Chinese Texts in Tibetan Transcription," BSOAS 21, 1/3, (1958), pp. 334-343):
     -經 kye
     -明/名 mye
     -相/想 zo
     -向/香 ho
     -放 pho
   Although from these examples we might expect *a bo from 阿傍, I think that still it could be a transcription, and have to look into more. Off the top of my mind, I remember that the surname Zhāng 張 is rendered as in Khotanese documents. I am completely foreign to the Mngon par rtogs pa mu tig gi phreng.  Do you Allan think it could have been translated from Chinese?
   Two corollaries to my previous post inspired by Hucker's Dictionary of Official Titles in Imperial China:
   -To my hypothesis no.1: ājiàn 阿監 "Eunuch  Attendant  upon  the  female  Chief of  Palace  Surveillance  (kung-cheng),   apparently  assigned  from  the  Palace  Domestic  Service  (nei-shih  sheng)." Literally, though, it would seem to be only "Mr. (?, =ā 阿) Supervisor."
   -To my hypothesis no. 2: Browsing through Hucker, I realize that one of the most common structures of titles is a nominalized verb+object phrase (e.g. 占夢, 知府, 織錦, 解試), rather than a concatenation of two verbs. So perhaps if ē zhù 阿助 is "to comply/obey with aid," then perhaps ēbàng 阿傍 could be "[those who] comply/obey with assistance"? And perhaps "ājiàn 阿監" is actually ējiàn "[he who] complies/obeys with supervision"?



   I apologize, I'm geeking out on this. Corollary 3:
   How do you feel about this passage in the scene of the "exorcism" of Chen Qingzhi 陳慶之 in the Record of Buddhist Monasteries in Luoyang?:
       自呼阿儂語則阿傍 (T51.2092.1009b14-15)
Yang Yuanshen 楊元慎 in his role of "exorcist" is trying to humilliate the southerner Qingzhi. Could this be a reference to Qingzhi's earlier appointment as shìzhōng 侍中 "palace attendant" with a pun like "You refer to yourself [with the southern pronoun] ānóng 阿儂 (*"I"), [but what we] call [you] is actually *guardian of hell (ā/ébàng 阿傍)"? Or perhaps "[... what we] call [you] is actually *lowly administrative assistant" making a reference to what could have been an actual bureaucratic title?
   I don't have here the recent French translation, and I don't recall if what they do with this passage is along those lines but Wang Yi-t'ung's "you call yourself A-nung and call others 'so and so' " does not make good sense to me. If this passage has been convincingly solved and I am not aware of it, I would be glad to whoever could enlighten me.




Dear friends,

I am unfortunately not near my library and so cannot check, but I would suggest looking at Daniel Berounsky’s édition and translation of the Tibetan Scripture of the Ten Kings. If it’s there, then I would think it to be most likely Chinese and not Indic in origin, a-bang thus being probably pseudo-Sanskrit as sometimes occurs in Chinese works.


Matthew Kapstein

Hi Eric,
I'm just back from Japan and jet-lagged, so can only give a rudimentary response. I tried to find a Chinese text using 阿傍(獄卒) Ābang, hell guardian that had an available Sanskrit version against which to compare, and gave up before exhausting possibilities, so such a text may be available, but I didn't identify it with a partial search. What I did find was 44 appearances of 阿傍 in the canon, more than a few in early Chinese translations which, despite unavailable Skt, are deemed genuine translations. Ābang seems to be the name of a specific hell guardian, identified, as you know, as Ox-headed in some of them, rather than a generic type of hell guardian. Since it is a specific name, not a type, identifying an underlying Indic original is problematic without finding a specific text that discusses him. Chinese transcriptions of Indian names and terms are frequently truncated, and since Ābang is only two characters, the probability of the name being truncated is great.

Here are some sample passages from relatively early texts in which the name appears:

《弊魔試目連經》卷1:「「時魔波旬,墮大地獄苦痛無量,時泥梨傍往語之言:『子欲知之,若有一籌,一鳥飛現,知過十千萬歲,如是之比亦復難限。弊魔!吾在地獄壽數如是,然後乃從大地獄出,更復遭厄二萬餘歲。』爾時,弊魔甚大愁毒。」」(CBETA, T01, no. 67, p. 868, a20-24)
tr by Zhiqian, roughly 2nd quarter of 3rd c. CE.

《大方便佛報恩經》卷2〈4 發菩提心品〉:「佛言:「善男子!過去久遠不可計劫生死中,時以重煩惱,起身口意業故,墮在八大地獄,所謂:阿訶訶地獄、阿婆婆地獄、阿達多地獄、銅釜大銅釜、黑石大黑石,乃至火車地獄。我於爾時,墮在火車地獄中,共兩人竝挽火車。牛頭阿傍在車上坐,[1]緘[2]脣切齒,張目吹火,口眼耳鼻,煙炎俱起,身體[3]殊大,臂脚[4]盤結,其色赤黑,手執鐵杖,隨而鞭之。我時苦痛,努力挽車,力勵前進。時我徒伴劣弱少力,劣弱在後。」(CBETA, T03, no. 156, p. 136, a8-17)
《大方便佛報恩經》卷2〈4 發菩提心品〉:「雖作如是唱喚,無益於己。我時見是,受大苦惱,心生哀愍,因慈心生故,發菩提心。為此眾罪人故,勸請牛頭阿傍:『此罪人者,甚可憐愍。小復加哀,垂慈憐愍。』牛頭阿傍聞已,心生瞋恚,尋以鐵叉前刺我頸,尋時命終,即得脫於火車地獄百劫[5]中罪。我以發阿耨多羅三藐三菩提心故,即脫火車地獄之罪。」」(CBETA, T03, no. 156, p. 136, a20-26)
《大方便佛報恩經》卷4〈6 惡友品〉:「爾時阿難受如來教,至地獄門外,問牛頭阿傍言:「為我喚提婆達多。」
阿難言:「我喚釋迦牟尼佛提婆達多。」爾時[8]牛頭阿傍即語提婆達多:「阿難在外,欲得相見。」」(CBETA, T03, no. 156, p. 148, b13-18)
Translator unknown, translated sometime during the Latter Han (1st - early 3rd c CE). 阿傍 by name appears 7 times in this text.

《出曜經》卷10〈9 誹謗品〉:「爾時世尊告諸比丘:「瞿波利比丘者,自招禍患入尼羅浮地獄中。爾時獄卒阿傍以鐵剛鉗拔出其舌長數百丈,」(CBETA, T04, no. 212, p. 665, c4-6)
“… at that time the hell guardian Ābang took hard iron pincers and pulled out [the victim’s] tongue many hundreds of feet.”
The Dharmapada/Udānavarga tr. by Zhu Fonian in 374 CE. Unfortunately this is part of the prose commentary by Dharmatrāta; while Pali and some Sanskrit versions of the verses are available, Dharmatrāta’s comm. is not.

《最勝問菩薩十住除垢斷結經》卷3〈9 定意品〉:「魔復思惟:『我有僕使名曰阿傍,亦能現變威德無量,當生之日天地大動,若取命終地亦大動,或能即是現瑞怪耳。』」(CBETA, T10, no. 309, p. 986, b20-23)
Also tr. by Zhu Fonian, ca. 365-385 CE.

《菩薩投身飴餓虎起塔因緣經》卷1:「「於時,仙師呪願太子因為說法。太子心喜,志樂無為,不欲還國,顧惟宮室生[16]地獄想,妻子、眷屬生杻械想,觀五欲樂為地獄想。思惟是已,即解瓔珞嚴身上服及車馬人從,悉付傍臣遣令還國。於是太子[17]披鹿皮衣留住山中,從師學道,[18]攢尋道術。」(CBETA, T03, no. 172, p. 426, b5-11)
Translation by Fasheng, ca. 397-439 CE

《賢愚經》卷1〈5 海神難問船人品〉:「賢者對曰:「更有可畏,劇汝數倍。」海神復問:「何者是耶?」答曰:「世有愚人,作諸不善,殺生盜竊婬[15]妷無度,妄言兩舌惡口綺語、貪欲瞋恚,沒在邪見;死入地獄,受苦萬端。獄卒阿傍,取諸罪人,種種治之,或以刀斫,或以車裂,分壞其身,作數千段。或復臼擣,或復磨之,刀山劍樹,火車鑊湯,寒水沸屎,一切備受。荷如此苦,經數千萬歲,此之可畏,劇汝甚多。」」(CBETA, T04, no. 202, p. 354, c3-11)
《賢愚經》卷13〈58 蘇曼女十子品〉:「當入地獄,受諸苦惱,數千萬歲,常為鹿頭羊頭兔頭諸禽獸頭阿傍獄卒之所獵射,無[4]央數歲,雖思解脫。」(CBETA, T04, no. 202, p. 441, a14-16)
Damamūka (nidāna sūtra), Translated by Huijiao in 445 CE. (there is a Tibetan version མདོ་མཛངས་བླུན་ as well as Mongolian tr. from Tib., but all seem to have originated in China from Chinese monks who heard these stories in Khotan, so no Sanskrit version to check against — these are Jātaka tales).
Cf. https://www.rigpawiki.org/index.php?title=Sutra_of_the_Wise_and_the_Foolish

So, assuming all or most of the texts cited above are translations from Indian originals, and that would be a reasonable assumption, some hell guardian was named Ā-something in Indian buddhist legend literature (early transcriptions did not reflect much later Mandarin pronunciations, so looking for a bang/pang is likely a red herring).

Does the name ring any bells with our Jātaka or Avadāna specialist colleagues?

You are certainly aware that the term 傍生 was used for being reborn in the animal realm; e.g.:
《大乘本生心地觀經》卷6〈6 離世間品〉:「觀畜生道而為恐怖,傍生界中受愚癡報,經無量劫難出離故。」(CBETA, T03, no. 159, p. 319, a11-12).

I hope some of that helped.


One additional thought: If, instead of trying to find a Sanskrit parallel text to one of the Chinese texts in which the name appears, one tries to imagine a possible Sanskrit term that resembles ā-ping, and taking into account that ā-ping is likely a truncated form of the underlying Indic name, AND keeping in mind that 牛 niu doesn't only mean "ox," but can also mean "bull" or "cow," there is a candidate. Something like āpiṅgalaka.

Monier-Williams defines that term: "ā-piṅgalaka m. a bull which has been set at liberty."

It derives from āpiṅgala, which means "reddish eyes," perhaps a moniker appropriate for a hell being.

Just a thought.


Dear Colleagues,
While Dan's contribution is of course very welcome, I wonder whether as a field we cannot do a bit better job at taking into account the fruits of scholarship. Just to cite a couple of examples, Dan refers to 弊魔試目連經, and credits it to Zhi Qian (note that this is correct, not Zhiqian). However, a look at the invaluable resource noted on this very list (https://networks.h-net.org/node/6060/discussions/837287/resource-online-...), this attribution is considered highly questionable by experts: https://dazangthings.nz/cbc/text/861/.
Furthermore, Dan cites a text as follows: "Damamūka (nidāna sūtra), Translated by Huijiao in 445 CE. (there is a Tibetan version མདོ་མཛངས་བླུན་ as well as Mongolian tr. from Tib., but all seem to have originated in China from Chinese monks who heard these stories in Khotan, so no Sanskrit version to check against — these are Jātaka tales)." I am afraid that there are a number of dubious statements here, beginning with the imaginary Sanskrit title, continuing to the citation of the Tibetan title, the name of the "translator" and the connection with Khotan just as a start.
That this type of misinformation continues to infect the field is a problem, I believe, that a small amount of care would go far toward eliminating.
Another perhaps trivial note: Diego cited an article by "Walter," but unless Diego is on a first name basis with (the late) Walter Simon, this is not quite apposite. The scholar Walter Simon should be better known than he is, and this relative obscurity perhaps contributed to this easily correctable oversight.

Best, Jonathan

Dear Dan, and all,

> „assuming all or most of the texts cited above are translations from Indian originals, and that would be a reasonable assumption…”

I am afraid I must disagree. All of the texts you cite are problematic in some way. My database is designed precisely to help with such questions.

T156 is a well documented case of a Chinese composition.


Nattier (2010) showed that T309 at least includes significant portions that were also composed on the basis of earlier Chinese sources.


As you say, Dan, T202, as shown by studies by Mair and others, is not really an Indic text.


Nattier does not regard the ascription of T67 to Zhi Qian as reliable, and scholars as far back as Sakaino and Hayashiya had already discussed problems with the ascription. This casts doubt on the date, at least, if not the provenance of the content.


Hiraoka has argued that unparalleled parts of T202 may be of Chinese composition, which would make the text as a whole, in Funayama sensei’s terms, “between translation and composition”.


The nature of T172, including the question of whether it is a real translation, is unclear (at least to me), but Hayashiya argued that the ascription to Fasheng, at least, is problematic.


Best wishes,

Michael Radich

My apologies to the memory of the late Walter Simon. Please correct: Simon, Walter. "A Note on Chinese Texts in Tibetan Transcription," BSOAS 21, 1/3, (1958), pp. 334-343.

Thank you Jonathan for the opportune remark.

Dear colleagues,
Epang 阿旁 and epang (or ebang) 阿傍 are regarded in dictionaries as transliterations. I doubt it. To derive it from āpiṅgala is far-fetched, as Indian piṅ(g) was transcribed by 賓, 并, 氷, while 傍 was used to transcribe vaṅ(g), though theoretically bhaṅ, phaṅ or paṅ could stand for 旁/傍.
In these compounds,阿 (“close, near; follows, relies”) and 旁/傍 (“side, near; helps, assists”) seem to be used synonymously. I assume that 阿旁/阿傍 may mean “a follower; a subordinate” or “one, who assists from the side; an assistant”, corresponding to BHS. upasthāyaka, BHS. upasthāka, Pā upaṭṭhāka, meaning “an attendant, a servant”, deriving from upa-√sthā (“to stand near in order to serve, attend, serve”).
This interpretation fits the case of 《最勝問菩薩十住除垢斷結經》魔復思惟:『我有僕使名曰阿傍,……』」(T.10, no. 309, 986b20f.; “Māra thought : ‘I have a servant, namely 阿傍 …..’ ”). However, I must admit that the word upasthāyaka does not fit very well the compounds 獄卒阿傍 / 阿傍獄卒 (“guardians of a hell and servants [of a hell]”?).

With best regards,
Seishi Karashima

Dear Prof. Kapstein,
I just found Berounsky's book, and unfortunately the Tibetan translation of the Scripture on the Ten Kings does not include the name of the ox-headed one (somehow a critical edition is not provided by this very interesting book). Here are the Chinese and Tibetan,
[17a] / bdun tshig gnyis pa'i nyi ma la / gshin rje hung wang zhes bya ba'i nyi ma la phyin no // de'i tshigs su bcad pa smras pa la / bdun tshig gnyis pa'i nyi ma la rgya tsha'i (read mtsho'i) rlung las rgal / thum po stong phrag khri khrag 'khrol ba'i chu'i nang du khyer / glang mgo can gyi srin pos ber rga (read rgyug) khur nas sngon du 'gro / rta mgo can gyi rag shas rtse gsum kha khyer nas 'jug tu bskul /

Informed by the discussions of the Chinese texts, I checked the Tibetan translation of the 大方便佛報恩經 (Thabs mkhas pa chen po sangs rgyas drin lan bsab pa'i mdo, Derge 353), and it translates 牛頭阿傍 into dmyal ba'i srung ma a bang ("Narakapāla A-bang"). But again, A-bang only shows up in Derge 353 in the whole Tibetan canon. The Tibetan translation of the 賢愚經 (mDzangs blun mdo) seems to simply skip this proper name. It seems that you are right that the Tibetan transliteration A-ba(ng) is very likely from a Chinese source.

The famed tertön Sprul sku tshul khrims bzang po has this passage in his rDzogs chen dgongs pa zang thal gyi khrid yig:

las 'bras mun sel las / ... 'chi khar 'gyod pa cung zad skyes te shi song / lhan cig skyes pa'i lha bu pho 'dre rgyal / chos rgyal kyis las ngan sbyang phyir sdug bsngal myong bar gyis gsungs / ... a ba glang mgos ded tshe / nga rgyal las kyi rnam smin gyis / dmyal srung a ba glang mgo can lag na mkhar ba'i be con thogs / rgyab nas 'phul zhing skrod par byed /

Unfortunately, Stéphane Arguillère, the French translator of this text, has a note saying that he does not what this terma titled "Las 'bras mun sel" is. He translate "A ba glang mgo" into "Enfant-à-tête-de-bœuf" (I don't know why he chooses "enfant," because even we take "a-ba" as "a-pa", it would be "father").

Best wishes,

Dear Colleagues,

阿傍 is a name of a ghost, servant of Bull Head (Gośīrṣa 牛头), per the following examples:
1) 《五苦章句经》http://tripitaka.cbeta.org/zh-cn/T17n0741_001
2) 《铁城泥犁经》 http://tripitaka.cbeta.org/zh-cn/T01n0042_001

This likely originated from a Chinese folk religion and is not of Indian origin.

Choong Mun-keat
University of New England, Australia.

Dear Jonathan and Michael,

Lesson 1: Don't post when jet-lagged.

There is a certain irony on being called out for lack of attention to the authenticity of supposed translations, since that has been one of the things I have been critical of the field for its naive acceptance of traditional attributions that get repeated uncritically in commonly used secondary and reference literature. Thanks for calling me out on this. That shows the field is improving. Jonathan, you are right, we should do better. I usually try harder.

You are both completely correct -- those texts have all been flagged as problematic "translations," and Michael's website https://dazangthings.nz/cbc/ is an invaluable resource for gaining a quick overview of the scholarly consensus or varying positions on things like the authenticity of texts. I consult it frequently, and have strongly recommended it to students since it first appeared. Resources such as this need to lead to revising or replacing the standard naive reference works still used by most of us.

As I mentioned, I was jet-lagged when I wrote that message, and taking shortcuts. Nevertheless, I was indeed aware of most of those texts' problematic nature, and Eric and I have discussed this offline. Just to add to the call-out, T212, the Dharmatrāta comm. to the Dharmapada, contains numerous sinitic elements, such as frequent mentions of Chinese geographic locations (Mt. Tai, etc.), so its Indic status is questionable as well, though in general it reads like a genuine Indian text.

By listing T202 last, I was signaling, perhaps too obscurely, the problematic nature of these ascriptions (all taken from the Lancaster, et al. catalogue of the Korean Canon, which frequently provides the traditional, naive attributions -- in need of revision), but in this case, with an additional link to a site that gave the account I repeated, an account which precisely undermined the reliability of that text as an authentic translation.

The mdzangs blun zhes bya ba'i mdo མཛངས་བླུན་ཞེས་བྱ་བའི་མདོ། , for which the AIBS database gives the Sanskrit title damamūko-nāma-sūtra [haven’t we realized yet that the “nāma” in transcribed titles in Tibetan are quotation marks comparable to iti, and not part of the actual titles?], and which others, such as the Lancaster catalogue give as Damamūko [nidāna sūtra], is, even according to the legend given on the rigpawiki site, not a translation of an actual text, but supposedly a Chinese rendition of what was heard by Chinese monks in Khotan. That legend, likely spurious, however may be a clue to how a name like ā-ping came to be, viz., oral accounts received by a Chinese audience, in which case the name may have an Indic origin after all. In addition to the Chinese, Tibetan and Mongolian versions (attesting to its continued appeal), there is also apparently a Tocharian fragment: Cf. https://gallica.bnf.fr/ark:/12148/bpt6k93307s.image.f333.pagination.langEN

What did strike me as curious was that a variety of texts from different sources were all adopting the same name. Further, while the texts have been flagged by the scholarly community, many seem to contain large amounts of authentic Indic material, suggesting they may represent a category that Toru Funayama has discussed, namely hybrid texts, i.e., texts written or compiled in China that contain strong indic input. If so, that adds some uncertainty to deciding which specific elements in such texts are themselves authentic or not. Things are not all black and white. In short, there are clear pseudepigraphic (a.k.a. "apocryphal") texts (the Awakening of Faith, the longer Śūraṅgama Sūtra, etc.), many of which have remained among the most popular Buddhist texts in East Asia for over a millennium; there are authentic Chinese translations of Indic texts; and there are a range of hybrids, that have murkier origins.

So, yes, mea culpa, it isn't "reasonable" to conclude that the texts I listed are authentic translations. That doesn't automatically disqualify 阿傍 as an attempt to render an Indic name, however. That it doesn't appear in bona fide translations does raise legitimate suspicions, buttressed by its appearance in apocryphal and hybrid texts (e.g. T620, see 339b5-8; cf. https://dazangthings.nz/cbc/text/2067/ as another appearance in a hybrid text), and eventually adopted in East Asian compositions.

As for the name 阿傍 being an attempted translation for "side helper," I'm afraid I don't find Seishi Karashima's suggestion persuasive. Not only because it is treated as a name, not a title or description -- it commonly appears as either 阿傍獄卒 "hell guardian ā-pang" and 牛頭阿傍 "ox/bull-headed ā-pang" -- but because pronunciations of Indic terms were constantly in flux, as were attempted transcriptions over time. If one reads the Chinese travelogue literature (Faxian, Xuanzang, Yijing, et al.), one of the distinctive features is that each succeeding travelogue claims its predecessors' transcriptions of place names and personal names were wrong, and then offers its own transcription, few if any sounding accurate in modern mandarin. None of those authors seemed to have a sense of how pronunciations of Chinese characters change over time, or even allowing for regional variations, as if assuming how they themselves pronounce a character is how the character was always pronounced everywhere. Kūkai, in his Heart sutra commentary, when comparing the various translations available to him, dismisses Xuanzang's translation on the grounds that the pronunciation of the mantra at the end is ineffective. Additionally the Indic source material -- whether written or oral -- was not always pure Sanskrit, but especially in the earlier periods, prakrits of various sorts. Equivalence charts are guidelines and suggestions, not restrictive parameters, as open to revision and addition as Kalgren's initial attempts. The interchangeability of 旁/傍 is indicative of their homophonic relation (like 相/想), not their semantic reference, indicating it is the phonetic aspect, not the semantic aspect that was at play, buttressed by 阿 as a common indicator of a transcription beginning with ā- or a.

Whether of Indian, Chinese, or hybrid origins, a study of the history of these hell guardians would be helpful. By the early Tang, Daoxuan’s 道宣 Guang hong ming ji 《廣弘明集》 contains a section titled “Descending to the dark depths of the hells” (沈冥地獄門) T.52.2103.311a25-c6, in which he addresses the skepticism of those who doubt the existence of the hells and the hell guardians. The Guang hong ming ji draws on earlier sources. Those who are skeptical have failed to look carefully and examine the evidence, and need to read the scriptures and treatises more carefully, Daoxuan argues. He says there are major hells, and a variety of minor hells scattered around the Iron mountain, in which denizens experience hot or cold, etc. and a variety of tortures. He says:

"Each of the hells has a leader, an ox/bull-headed ā-ping, whose nature is to cruelly torture without an ounce of mercy. He watches others receive pain; he only observes their suffering, but feels no pain himself; he only observes their tortures, but remains unscathed himself."

[yes, I am glossing 毒 as torture, not poison, no need to "correct" me on that.]

When these hell guardians are questioned why they don't suffer the tortures they inflict, they explain that the ones who suffer have committed the five grave transgressions, or all manner of misdeeds, killing, stealing, debauchery, addictions to alcohol and food, criminal activities, etc., and were unrepentant. The punishments inflicted are deserved, designed to turn them from their ways to instead seek liberation and nirvana (and thus are really compassionate!).

It is in the bhāṣya to AK 3:59ac that Vasubandhu raises the question of whether the hell guardians are sentient beings or not (kiṃ te narakapālāḥ sattvasaṃkhyātā utāho neti?). He entertains both possibilities, fending off objections to both options.

If they are not actual beings, how do they move, what animates them? They are animated by wind from the karmic seeds of actual sentient beings. (kathamidānīṃ ceṣṭante? sattvānāṃ karmabhirvivarttanīvāyuvat). That’s a clear predecessor to the position Vasubandhu takes in the Viṁśikā. How then, to account for a verse by the Bhadanta Dharmasubhūti:

"krodhanāḥ krūrakarmāṇaḥ pāpābhirucayaśca ye |
duḥkhiteṣu ca nandanti jāyante yamarākṣasā"

Paramārtha: 恒瞋最麁業 於惡起愛樂 見他苦生樂 必作閻摩卒
Xuanzang: 心常懷忿毒 好集諸惡業 見他苦欣悅 死作琰魔卒
Buddhavarman: 剛強瞋恚人 常樂作諸惡 見他苦生喜 死作閻羅卒

which is cited in abhidharma Vibhāṣa texts as confirmation that hell guardians are real sentient beings subject to karmic retribution (e.g. Buddhavarman's tr. of a Vibhāṣa in the 5th c, T.28.1546.48a5-25, esp. a9-10, quoted above; Cf. also Mahāvibhāṣa, tr. by Xuanzang: T.27.1545.866b16-17)). Pruden renders la Vallée Poussin’s French translation thus: "Those who are angry, who take pleasure in cruel actions and transgressions, who rejoice in the sufferings of others, are reborn as Yamarākṣasās?” Vasubandhu replies that they do Yama’s bidding, but are not agents of their own actions. (ye te yamenānuśiṣṭāḥ sattvān narakeṣu prakṣipanti ta ete yamarākṣasā uktā na tu ye kāraṇāḥ kārayantīti).

If they are sentient beings, where do they suffer retribution for their wicked actions? In the same hells in which they inflict them, is the answer. So the jailers become the jailed in turn. Why are they not harmed by fire, etc.? Because their karmic power prevents the fire from reaching them, or they are composed of special elements.

Sanskrit: Pradhan p. 164 lines 13-23
French: la Vallée Poussin, v. 2, pp. 152ff
English: Pruden, v.2, pp. 458f

getting sleepy...

Well, this has been a stimulating discussion. Since I began this conversation, I had better get a few words in.

1. For the record, the citation from the 五苦章句經 (T.741) that was recently posted contains an error. The actual line reads 獄卒名[+var 阿]傍, not 鬼將名阿傍 (which would suggest a distinctly "Chinese folk religion" kind of origin, were this in fact what the text said). This erroneous citation of T.741 has somehow found its way onto Chinese Wikipedia (https://zh.wikipedia.org/wiki/牛頭馬面) and, now, no doubt other sources. This text contains, here at the relevant section, a parallel to the Devaduta-sutta of the Majjhima-nikāya. (More on this in a second)

2. What has been most interesting for me in this conversation is the discovery of what I am now taking to be the “original” form of 阿傍, namely 泥梨[犁/黎etc.]旁[傍], without the character a/e 阿. After seeing this word in T.67 first cited by Lusthaus (who is forgiven his mistaken translator identifications, by me at least, for having pulled this text up), I noted it elsewhere: T.23, T.42, and T.86. Without going into the details, these texts are all of somewhat unclear background (their translator attributions are debatable), and some of them seem interrelated (i.e., perhaps copied from each other in parts). But they are all, I think, “early” – pre late fourth century and hence much earlier than any example of the classic form 阿傍[旁], that I can see at least.

3. Two of these (Alan [Yi] Ding found one of these) have clearly identifiable Indic parallels

a. T.67, in the Māratajjaniya-sutta , where 泥梨傍 is parallel to nirayapāla.
b. T.42, parallel to the Devaduta-sutta, and again 泥犁旁 = nirayapāla

4. Now that we can leave aside 阿 as somehow a latter addition, and we have two nominally independent examples of it, could泥梨傍 be a phonetic rendering of (insert suitable Middle Indic form of) niray-pāl[a]? Diego Loukota suggested to me off list that this is unlikely. Any other opinions? Obviously the word niraya-pāla is exactly the right meaning. Or is this one of those pseudo-translation/transliteration cases where the word is a translation (“follower” “attendant”) chosen because it almost sounds like the phonetic rendering?

5. Whatever it began as, the fact that the (few) Tibetan translations from Chinese don’t seem to know what to do with it, and that medieval Chinese lexicographical works (翻梵語, 一切經音譯, etc.) seem to NOT ever pick up this word, suggests that it at the very least came to be read semantically by medieval Chinese Buddhists.

That’s my summary of what I’ve learned anyway. Thanks everyone for your input.


Dear friends,

Just a few further notes on the Tibetan angle. So far as I have determined so far, the phrase a ba glang mgo is not actually used in the Tibetan Mdzangs blun, of which the "Sanskrit" title - Dāmomūrkhasūtra - is almost certainly not original, but a back translation from one or another of the languages in which the text was circulated. Tibetan gives only glang mgo, "ox head."

I have so far located just two Tibetan canonical texts that use the phrase a ba glang mgo. Both are tantric: Bdud rtsi thig pa, in vol. 43 of the Dpe bsdur ma ed. of the Tanjur, is attributed to Mtsho skyes rdo rje, *Saroruhavajra, often taken as a variant of Padmasambhava. This work is almost certainly not of Indian origin.

Mngon par rtogs pa mu tig gi phreng ba, in vol. 47 of the Dpe bsdur ma, is attributed to Mitrayogin, an Indian teacher active in Tibet during the 11th c. Many of his works, including this one, appear to have been composed in Tibet, however.

In sum, Tibetan canonical sources give us no clear evidence of an Indic background of a-bang.

As for the French translator's rendering of "enfant," mentioned by Yi Ding, this may be explained as an attempt, given the absence of any clear definition, to see a-ba as a possible spelling of the colloquial Tibetan awa, "child." When I've seen this written, however, the gsal byed wa and not ba is usually used.

Finally, although I have not yet seen it in the Tibetan Mulian Jiumu 目连救母 traditions with which I am familiar, this cannot be ruled out as a source. Several Tibetan versions of the bianwen are known.



Matthew Kapstein