Reading the Sixteen Arhats: how the jūroku rakan are named, placed, and told apart

The Sixteen Arhats are a closed, named roster fixed by the Nandimitra Record, translated by Xuanzang in 654: sixteen disciples charged to remain until Maitreya, each with a set abode and a retinue count. Reading a set is positional and named, not generic — find the scroll’s number in the sixteen, match the figure to the roster attribute, and watch where China’s slide to eighteen did not follow into Japanese Zen, which mostly kept the sixteen.

The set is the same set every time

A five-hundred-arhat set is a crowd; its reading is statistical, and a single surviving scroll is a fragment of a count. The sixteen are the opposite case. They are a closed list of named individuals, and the reading is not “an arhat” but “the second one, Kanakavatsa, the one with the rope, who lives in Kashmir with five hundred followers.” The companion study on the five-hundred sets handles the crowd and its donor economy; this one stays with the named sixteen and the one number that changed.

Look at the Met scroll first. A single elder sits on a rock under a tree that has been drawn older than he is, robes piled in heavy folds whose outline is worked harder than the face inside them, an attendant at his side. Nothing on the silk says which arhat this is. Hung as the museum hangs it, alone on a wall, it reads as a generic luohan portrait; hung as it was meant to hang, second from one end of a row of sixteen scrolls spaced along a hall, the same painting is a specific named person and the row to its left and right tells you so. The Met catalogues it as Satsubari, the Second of the Sixteen Arhats, and the only reason that title can be written down is that the scroll once hung as panel two of a set of sixteen. Take it out of the set and the name goes with the wall. That dependence — face says elder, position says Kanakavatsa — is the whole method.

The roster was fixed by one text

The list is not folklore that drifted into shape. It was set by a single source: the Record of the Duration of the Law as Spoken by the Great Arhat Nandimitra (大阿羅漢難提蜜多羅所説法住記, Sanskrit Nandimitrāvadāna, commonly the Fazhuji), translated into Chinese by Xuanzang in 654. Mecsi calls it “the most important and most cited source,” the text that “names and introduces a group of sixteen arhats according to their realms” and explains “how Buddha entrusted the arhats with the protection of Buddhist teachings.”¹ The frame is the one that gives the genre its point: these sixteen were charged not to enter final extinction but to remain in the world, holding the teaching intact, until the future Buddha Maitreya arrives. They are the disciples who did not leave.

What the roster is not is a list of the Buddha’s closest pupils. Mecsi, following de Visser, notes that of the Buddha’s ten chief disciples only one (Rāhula) appears among the sixteen, and that of the twenty-nine named in the five-hundred group only four overlap with the sixteen (Piṇḍola Bhāradvāja, Vakula, Cūḍapanthaka, Rāhula).¹ The sixteen are a separate construction with their own logic: Mecsi reads the number as “a multiplication of the four directions to protect Buddhist teachings.”¹ Four guardians per quarter, the teaching held at the compass.

Each name carries an address and a headcount

The Nandimitra roster does not just name the sixteen. It assigns each one a fixed dwelling-place and a fixed number of attendant arhats, and the counts climb. Reading a labelled set means knowing that the cartouche is checkable against this table, and that an unlabelled set can sometimes be back-solved from it:

#	Sanskrit	Abode (per Nandimitra)	Retinue
1	Piṇḍola Bhāradvāja	Godhānya / western continent	1,000
2	Kanakavatsa	Kashmir	500
3	Kanaka Bhāradvāja	Pūrvavideha (east)	600
4	Subinda	Kuru country	800
5	Nakula (Bakula)	Jambudvīpa	800
6	Bhadra	Tāmradvīpa (Ceylon)	900
7	Kālika	Siṃhala (lion country)	1,000
8	Vajraputra	Parṇadvīpa	1,100
9	Gobaka (Jīvaka)	Gandhamādana	900
10	Panthaka	Trāyastriṃśa heaven	1,300
11	Rāhula	Priyaṅgudvīpa	1,100
12	Nāgasena	Pāṇḍava mountain, Magadha	1,200
13	Aṅgaja	Vipulapārśva	1,300
14	Vanavāsin	the habitable mountain	1,400
15	Ajita	Gṛdhrakūṭa (Vulture Peak)	1,500
16	Cūḍapanthaka	Iṣadhara, near Sumeru	1,600

The retinue figures total roughly seventeen thousand and are themselves part of the iconographic record: a labelled scroll will sometimes give the abode and the number in the cartouche, and a set that has lost its labels can occasionally be reordered by reading them back out of the painted scene. The numbers are doctrinal furniture, not decoration. They are why the genre insists, against the eye, that this seated elder is the second and not merely an elder.

Read the attribute, but trust the number

The roster carries conventional attributes, and a reader who knows them can often name a figure on sight. The handbook shorthands run: Piṇḍola with long white eyebrows and a book or alms-bowl; Kanakavatsa with a jewelled rope or lasso held in both hands; Kālika with a heavy gold earring worked in each hand; Nāgasena with a vase and a monk’s staff; Rāhula crowned or jewel-headed; Bakula (Nakula) with a mongoose; Aṅgaja with a fly-whisk and an incense bowl; the dragon-and-bowl motif attaching to Panthaka in several lists.²

The Met Satsubari is the working example of why the attribute is the second move and not the first. Kanakavatsa’s roster sign is the jewelled rope; the Burke catalogue records this scroll as Satsubari, the Second of the Sixteen Rakan, so the painter was working to the second slot.³ But the rope is a small thing in a large silk, and the figure that carries the reading is an old man on a rock who would, lifted out of his set, be indistinguishable from the eighth or the fourteenth. The Guanxiu tradition pushed this further. Guanxiu, the monk-painter (832–912), “took his inspiration from his dreams when depicting the extraordinary-looking arhats with exaggerated features as sitting under trees or meditation in caves,” Mecsi notes; the faces were driven toward bushy-browed, slack-jawed, foreign exaggeration precisely because they were never going to be individuating.¹ The faces are types. The number is the identity. Read the cartouche, count the position in the hang, then let the attribute confirm what the number already told you, not the other way round.

The common misreading inverts this. A single arhat scroll in a museum gets captioned “an arhat” or “a luohan,” and the set-membership, the one fact that makes the figure a named person, is dropped because the panel arrived alone. The Met’s own title resists that: it keeps “the Second of the Sixteen” in the object name even though only the one scroll is held. The genre’s first instruction is to refuse the generic caption.

A second instruction follows from how these sets were physically used. A jūroku rakan set is not a polyptych with a fixed centre; it is sixteen independent hanging scrolls, stored rolled, brought out and hung in number-order for a rite (the rakan-ku, the offering to the arhats) and rolled away again. That use-history is why single scrolls survive scattered across collections while complete sets of sixteen are rare, and why the Met Satsubari is alone: it is one scroll from a dispersed set, its fifteen companions elsewhere or lost. The number in the title is the set reassembled on paper after the silk was scattered. Reading it well means reading the absent fifteen as much as the present one.

The set was sixteen before it was eighteen

The single most useful thing to know when reading an East Asian arhat set is where it sits on the sixteen-to-eighteen fault line, because the number is a dating-and-region signal. The Nandimitra roster is sixteen. Somewhere between the late Tang and the Five Dynasties, in China, two more were added to make eighteen, and the eighteen-figure set became the dominant Chinese form.⁴ The added pair is not stable across the tradition. The Sino-Tibetan stream, Mecsi notes, “sometimes” adds Dharmatrāta and the “Big-bellied Buddha,” the latter the pot-bellied Hotei/Budai figure.¹ Other Chinese lists name Nandimitra himself (the narrator of the source text, folded into the roster he reports) or Piṇḍola the Younger, or fill the two slots with the popular taming-the-dragon and taming-the-tiger arhats, which the Eighteen Arhats tradition records as later, partly anti-Daoist additions.⁴ The instability is the tell: the +2 are an accretion, and which two you get encodes which regional stream and which century the set descends from.

This is where the Japanese case diverges, and the divergence is the reading rule. The eighteen-figure expansion, in Mecsi’s account, “only gained a foothold in China, whereas other areas like Japan continued to revere only sixteen.”¹ A set that presents as a clean sixteen, in a Japanese context, is the conservative line holding the Nandimitra number against the Chinese drift. The Met Satsubari is a Nanbokuchō scroll, Japan, late fourteenth century, and it belongs to a set of sixteen, not eighteen. Counting the panels, or reading the highest cartouche number, is therefore not pedantry; it is the first diagnostic for where and when the set was made.

Why Zen kept the sixteen

The reception that carried the sixteen in Japan is the Chan/Zen one, and it split along a line worth reading off the object. Chan Buddhists, Mecsi writes, “had a great affinity for arhats”: the cult of the five hundred “was popular among Rinzai and Ōbaku Zen in Japan, while Sōtō Zen preferred the representation of the sixteen arhats.”¹ So the very choice between a sixteen-set and a five-hundred-set is, in a Japanese monastic context, a school signal as much as an iconographic one: Sōtō leaning to the named sixteen, Rinzai and Ōbaku to the uncountable crowd. The companion study handles the five-hundred side; what matters here is that the sixteen are the set Zen could use as a closed roll of named exemplars.

The deeper Chan move is that the arhat became the source pattern for other figures. Mecsi argues that Bodhidharma iconography “has its sources in earlier or contemporaneous arhat paintings, as if one of those figures had been selected and represented separately,” and that Bodhidharma “is believed to be a kind of arhat, so this similarity is unsurprising.”¹ The hooded, bearded, foreign-featured elder under a tree, the Guanxiu arhat type, is the same visual stock the half-length Bodhidharma portrait draws from. A reader who has learned to see the sixteen has also, without noticing, learned to read the Daruma portrait, which is why the sixteen-arhat genre sits at the centre of the Zen-painting reading and not at its edge. (The reception-history of Daruma as a subject is its own study; what the arhat set supplies is the visual grammar it borrows.)

Where this reading commits, and where it is open

bodhi commits to the positional-and-named reading: the sixteen are not a generic class but a fixed roster from one datable text, and the correct first move is the number, not the face. The scholarship supports this. De Visser’s 1923 collection of the early textual evidence, still the foundation, and Fong’s 1958 study of the painted tradition both treat the Nandimitra roster as the fixed armature against which any set is read; Mecsi 2018 is the recent synthesis followed here for the China–Korea–Japan comparison and the Bodhidharma link.¹⁵⁶

Two things are genuinely open. First, the abode-and-retinue table varies in its margins across recensions and later handbooks: the abodes are stable, the retinue counts and the spelling of the lesser names (Subinda/Suvinda, Gobaka/Jīvaka, Nakula/Bakula) drift, and a serious page-level reconciliation of the Nandimitra figures against de Visser’s collated texts and Fong’s plates is a future operator pass, not something settled here from secondary sources. Second, the identity of the +2 in any given eighteen-set is often genuinely undecidable from the object alone: the same scroll has been read as Nandimitra, as Piṇḍola the Younger, or as a dragon-tamer by different cataloguers, and where the cartouche is lost the question stays open rather than resolving to a default. The honest reading names the sixteen with confidence, flags the eighteenth and seventeenth as contested, and counts the panels before it trusts the faces. The residual uncertainty is not in the sixteen, which the Nandimitra text closed in 654 and the Japanese transmission largely kept closed; it is in the two that China kept adding, and in every set that has lost the cartouche which alone could say which two it meant.

Related

• The Five Hundred Arhats set — the crowd-and-count sibling to this named sixteen
• The Sixteen Benevolent Deities — a different sixteen, the Daihannya guardians, not the rakan
• Jūroku rakan (concept)
• Piṇḍola Bhāradvāja (figure)
• Zen (tradition)

Footnotes

1. Beatrix Mecsi, “The Arhats and Their Legacy in the Visual Arts of East Asia,” Hualin International Journal of Buddhist Studies 1.2 (2018): 131–142 — for the Nandimitra/Fazhuji as the primary roster source and its Maitreya-protection frame (pp. 133–34, where Mecsi names the text and Xuanzang’s 654 translation, following Fong, The Lohans, p. 35); for the disciple-overlap statistics and the four-directions reading of the number (p. 134, following de Visser, The Arhats, pp. 197–98); for Guanxiu (832–912) and the dream-derived exaggerated type (p. 135); for the Sino-Tibetan +2 (Dharmatrāta and the “Big-bellied Buddha,” p. 135, following de Visser pp. 62–78); for the eighteen-only-in-China / Japan-kept-sixteen claim and the Rinzai-Ōbaku-500 vs Sōtō-16 split (pp. 135–36); and for the Bodhidharma-from-arhat-pattern argument (pp. 136–38). Page ranges keyed to the Hualin PDF pagination; specific page-pins for de Visser and Fong are taken at the level Mecsi cites them and are flagged as un-collated against the source volumes. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹

2. Conventional attribute shorthands (jewelled rope for Kanakavatsa, mongoose for Bakula, fly-whisk and incense bowl for Aṅgaja, dragon-and-bowl for Panthaka, etc.) as compiled in the public Nandimitra-roster handbooks; treated here as the empirical handbook convention, not as sutra-derived, and not page-pinned to de Visser or Fong. ↩

3. Mary Griggs Burke Collection catalogue no. 25, Satsubari (薩婆利), the Second of the Sixteen Rakan (十六羅漢) — the provenance and set-membership record for Met 2015.300.6; the burkecollection.org catalogue page was unreachable on access (connection refused) and is cited at record level, with the Met collection record (objectID 53180, accession 2015.300.6, public domain (CC0), verified via the Met Collection API 2026-05-16) carrying the object facts. ↩

4. “Eighteen Arhats,” Wikipedia — for the late-Tang-to-Five-Dynasties dating of the +2 expansion, the competing candidates (Nandimitra, Piṇḍola the Younger, Mahākāśyapa, the dragon-taming and tiger-taming arhats), and the record of the dragon/tiger pair as later, partly anti-Daoist additions; reference-level corroboration of Mecsi’s account, not a substitute for it. ↩ ↩²

5. M. W. de Visser, The Arhats in China and Japan (Berlin: Oesterheld & Co., 1923) — the foundational collation of the early textual evidence for the sixteen and the variant +2; cited here at work level via Mecsi’s apparatus, not collated against the printed volume (operator page-pin pass needed). ↩

6. Wen C. Fong, The Lohans and a Bridge to Heaven (Washington: Freer Gallery of Art, Smithsonian Institution, 1958) — the standard study of the painted arhat tradition and the Daitoku-ji set; cited here for the Nandimitra/Xuanzang dating as relayed by Mecsi (Fong p. 35), at work level, not collated against the printed volume. ↩