The Binary Structure of Ifá: How 256 Odù Are Generated and Read

Dey dey write Odù as two column wey stand side by side, each column get four position wey arrange from up go down, and each position dey carry either one mark or two. Na only two option dey: position with one mark and position with two marks, wey make each position be binary digit. Four position per column dey give sixteen possible columns, two columns dey give sixteen times sixteen, and sixteen times sixteen na 256. Na all di arithmetic be dat, and every oda tin inside dis file na details about how dem dey produce di marks, how dem dey arrange di figure, and wetin di structure allow or no allow you to claim.

Bascom state di formula direct: each half of di figure dey take one out of sixteen basic forms "according to di formula 2^n for di number of permutations of head and tail for coin wey dem toss four times one after di oda, wey 2 na di number of possible alternatives (head or tail) and n na di number of tosses" 1 Den e give di combination: "total of 16 x 16 or 256 derivative figures (odu) dey possible" 1

Di mark, and wetin dem take make am

Di unit na mark wey dem draw inside wood powder on top tray, or seed shell wey face one side instead of di oda side. E get two states and no third one dey.

With ikin, di sixteen sacred palm nuts. Di diviner dey hold all di sixteen nuts for one hand and e go grab dem with di oda hand. Almost every time, some go remain behind. Di rule wey Abimbola give na say if one nut remain e go make two marks, and if two nuts remain e go make one mark 2. Bascom give di same rule and point out wetin strange about am: "If one nut remain, di diviner go draw two short parallel lines inside di divining powder for di tray; if two nuts remain, e go make single line. To justify dis reversal wey look like say na just anyhow, di diviners only talk say na so Ifa teach di Ifẹ̀ to do" 1

Make you note dis reversal, because e dey important for anybody wey dey try read di system as clean numeral notation. One nut dey give two marks; two nuts dey give one mark. Di connection between di physical event and di symbol wey dem record dey inverted, and dem no record any functional explanation for why e dey inverted. Di answer wey di tradition give na say na so dem teach am.

Di trial dey count only when na exactly one or two nuts remain. Bascom: "If none no remain, or if more dan two remain, or if di hold no firm and e feel say some dey 'try slip commot,' e go beat dem again and hold dem anoda time" 1 Abimbola give di same rule 2. Dis na rejection-sampling procedure. Outcomes wey dey outside di valid range, dem dey discard dem and repeat di draw, wey be real and specific piece of engineering: e dey convert physical process wey get plenty possible outcomes into clean two-state output.

Dem dey repeat dat procedure until eight valid marks don exist.

With ọ̀pẹ̀lẹ̀, di divining chain. Eight half seed shells, four on each side of chain wey dem hold for middle. Each shell dey land either make di concave side face up or di convex side face up. Bascom: half shell wey fall "open" with di concave inner surface facing up "na di same tin as single line for di tray; if e fall for di closed or 'inverted' position with di convex outer surface facing up, na di same tin as double mark" 1 Abimbola describe di same correspondence using di rough inner and smooth outer surfaces 2.

Di chain dey produce all di eight bits for one throw. Di palm nuts need eight separate manipulations, wey each one fit need several attempts. Bascom state di contrast plain: "Dem fit select one of di 256 figures with just one throw of di divining chain, but dem go need eight separate manipulations of di palm nuts" 1 Di two methods dey identical in terms of information. Bascom: "otherwise di two systems na di same. Dem dey use di same set of figures with di same names and rank order, and di same verses" 1

Orientation, and why e no be for decoration

Dem dey make di eight marks for fixed order, and dem no fit swap di two columns.

Bascom: di eight marks "dey arranged in two parallel columns of four marks each" and "di diviner dey work from right to left wen e dey mark di lines for di divining tray, wen e dey read di figure, and wen e dey choose between specific alternatives" 1 Abimbola describe di same tin: di diviner "dey make either one or two marks like dis four times for two rows starting from right to left" 2

So dem dey fill di right column first, from up go down, den di left column, from up go down.

Di right column get priority for naming. Bascom: "Each derivative figure dey get compound name based on di names of di patterns of di right and di left side. Di name of di right side dey come before dat of di left side" 1 E add di interpretation wey di tradition dey attach to am: di right (ọ̀tún) "dem dey take am as male and 'more powerful' pass di left (òsì), wey be female; na why di name of di right half dey come before di left" 1

Dis no be just matter of how dem write am. Bascom clear well-well say to reverse di columns go produce different Odù with different content: di figure wey dem show "na Ọ̀kànràn Ìrẹtẹ̀, and no be Ìrẹtẹ̀ Ọ̀kànràn. Because di second one na anoda figure, with different set of predictions and sacrifices, e dey very important make person differentiate between di two halves of di figure by making sure say di divining tray and di two halves of di divining chain face di correct direction" 1

Abimbola confirm di same convention from di naming side, wen e explain one minor Odù: "E real name na Ogbégúndá, wey mean combination of Ogbè for right and Ògúndá for left" 2

One practical consequence for people wey dey read secondary sources: plenty writings on top Ifá, including materials wey practitioners produce, dey present Odù diagrams without stating which column be which. Since na di ordered pair dey identify di Odù, diagram wey dem no label fit mean two different Odù unless say di figure symmetric.

Notation wey dem use for inside dis corpus

Sources get different ways wey dem dey take write the marks, so this corpus choose one standard and explain how e map am.

We dey write column from up go down as line of four numbers, using Bascom own notation from im Table 3 1:

• 1 = single mark (one stroke; ọ̀pẹ̀lẹ̀ shell concave side face up)
• 2 = double mark (two strokes; ọ̀pẹ̀lẹ̀ shell convex side face up)

Complete Odù dey written as RIGHT|LEFT, right column first, as e match the naming order. So Èjì Ogbè na 1111|1111 and Ogbèyèkú na 1111|2222, that is Ogbè dey the right side and Ọ̀yẹ̀kú dey the left side.

To convert am to normal binary, map single mark to 1 and double mark to 0, dey read the top position as the most significant bit. Na choice wey dem make for this corpus, no be wetin tradition specify, and the caution for the next section apply to am.

The sixteen columns, verified

Under Bascom notation, the sixteen named column patterns na these ones, and dem mathematically complete: all sixteen possible four-position patterns dey appear, each one exactly once, none miss and none duplicate. I verify this thing exhaustively against the sixteen-member set of four-digit strings over the alphabet {1,2}.

Column pattern (top to bottom)	Name	As binary (single=1)	Decimal
1111	Ogbè	1111	15
2222	Ọ̀yẹ̀kú	0000	0
2112	Ìwòrì	0110	6
1221	Òdí / Èdì	1001	9
1222	Ọ̀bàrà	1000	8
2221	Ọ̀kànràn	0001	1
1122	Ìrosùn	1100	12
2211	Òwónrín	0011	3
1112	Ògúndá	1110	14
2111	Ọ̀sá	0111	7
1121	Ìrẹtẹ̀	1101	13
1211	Òtúrá	1011	11
2212	Òtúrúpọ̀n	0010	2
2122	Ìká	0100	4
1212	Ọ̀ṣẹ́	1010	10
2121	Òfún	0101	5

The mark patterns wey dey this table na Bascom own, recorded from im Ifẹ̀ informants and cross-checked against im corpus of 86 lists 1. The binary and decimal columns na conversions wey dem do for this corpus under the mapping wey dem state above, and dem be arithmetic rather than ethnographic.

Why 256, and why not another number

Na three independent design choices force the number, and changing any one go change the total.

Two states per position, because the physical procedure dey bring exactly two outcomes: one nut or two, concave or convex. This one dey give base 2.

Four positions per column, wey be choice wey no mechanical thing force. Nothing for the palm-nut procedure say make person stop for four. Four positions dey give 2^4 = 16 columns. Three go give 8, five go give 32. The system stop for four, and sixteen dey repeat throughout Ifá: sixteen palm nuts, sixteen principal Odù, sixteen verses per Odù as the minimum for initiation, sixteen cowries for the related system. Whether sixteen come before the four positions or na consequence of dem no be wetin the sources settle.

Two columns, wey dey give 16^2 = 256. Two columns na wetin dey turn set of sixteen categories to set of 256 ordered pairs, and na wetin dey make the naming compound.

The order matter, na why the total na 256 and no be 136. If Ogbè-on-right-with-Ọ̀yẹ̀kú-on-left be the same Odù as Ọ̀yẹ̀kú-on-right-with-Ogbè-on-left, the number of distinct figures go be the number of unordered pairs with repetition, wey be (16 x 17)/2 = 136. Because Bascom document say the two na different figures with different content 1, the count na the full 16 x 16 = 256. The system explicitly non-commutative, and that non-commutativity roughly double im capacity.

Bascom correction of one old overclaim

Older literature inflate the number badly, and Bascom address am directly because e spread well well reach wetin need refuting.

E note say "many authors have claimed that the figures of Ifa number in the thousands," wey come from J. Johnson account of sixteen principal odù, "256 or 4,096 odù altogether, and, 'according to some' 65,536." Farrow inflate am further as e point to the many possible results of each of the eight drawings, wey Bascom observe say "is already accounted for in increasing the number from sixteen to 256." Im conclusion: "As any reader may verify for himself, however, the system permits no more" 1

The numbers 4,096 and 65,536 na 16^3 and 16^4, that is, wetin you go get if you put three or four columns instead of two. Dem arithmetically coherent but dem ethnographically wrong. The system get two columns.

This matter pass ordinary record keeping. The inflated figures still dey circulate, and dem dey sometimes bring dem as evidence say the system sophisticated. The accurate number na 256, and 256 don dey okay well well.

A structural property of the ranking

Real mathematical regularity dey for how dem arrange the sixteen principal Odù, and e hold for both of the major competing orders wey dem document for file 03. I check am exhaustively instead of taking am from a source, so e dey presented as arithmetic on Bascom recorded patterns rather than as ethnographic claim about intent.

Call column pattern vertically symmetric if e dey read the same top to bottom as bottom to top. Exactly four of the sixteen patterns get this property: 1111, 2222, 2112, 1221. Those four na precisely Ogbè, Ọ̀yẹ̀kú, Ìwòrì, and Òdí, wey be precisely the first four Odù for the ranking, for both the Ifẹ̀ and the southwestern orders wey Bascom tabulate 1.

The remaining twelve patterns form six mirror pairs, each pattern paired with im own top-to-bottom reversal. For both orders, every one of those six pairs dey occupy two adjacent positions for the ranking:

• Ọ̀bàrà 1222 with Ọ̀kànràn 2221
• Ìrosùn 1122 with Òwónrín 2211
• Ògúndá 1112 with Ọ̀sá 2111
• Ìrẹtẹ̀ 1121 with Òtúrá 1211
• Òtúrúpọ̀n 2212 with Ìká 2122
• Ọ̀ṣẹ́ 1212 with Òfún 2121

The two competing orders differ only for the sequence wey these pairs take appear and for which member of a pair dey come first. Their pairing structure dey identical.

Wetin this one show na say the way dem arrange the figures no be by chance when you look their geometry: symmetric figures come first, then the mirror twins dey together. Wetin e no show na the intention behind am. No source wey I see talk say na this principle dem take arrange am, and I no dey claim say na so e be. The pattern dey real and person fit check am; but the origin no dey documented.

Example wey dem work out, from cast reach signature

This one dey trace how the mechanics dey work on top cast wey dem just invent. Na illustration of notation, no be instruction for practice, and e comot the invocations, the ritual context, and everything wey dey make real consultation be wetin e be.

Make we assume say diviner dey work with palm nuts. Na the right column dem go first fill, from top go down, with four valid trial:

1. Two nuts remain, so na one mark. Position 1 for the right column: 1
2. Two nuts remain, so na one mark. Position 2: 1
3. One nut remain, so na two marks. Position 3: 2
4. One nut remain, so na two marks. Position 4: 2

The right column read 1122 from top go down, wey the table above call Ìrosùn.

Then the left column, from top go down again:

5. Two nuts remain, one mark: 1
6. One nut remain, two marks: 2
7. Two nuts remain, one mark: 1
8. One nut remain, two marks: 2

The left column read 1212, wey be Ọ̀ṣẹ́.

The full signature na 1122|1212. If dem name right column first, this one na Ìrosùn Ọ̀ṣẹ́. As eight bits under the mapping wey dey above, na 11001010, decimal 202, and as ordered pair of column values, na (12, 10).

If na the same eight marks appear but dem swap the columns, 1212|1122, the Odù for be Ọ̀ṣẹ́ Ìrosùn, wey be different figure with different verses. The eight physical marks na the same; na the orientation dey differentiate them.

The diviner go come recite the verses wey dey filed under Ìrosùn Ọ̀ṣẹ́ until the client stop am for the relevant one.

Wetin the binary structure support and wetin e no support

Na this section person suppose read carefully, because people dey often stretch the structural facts wey dey above pass wetin dem actually support.

Wetin dey established. Each position get two states for real. Four positions per column dey give sixteen columns, two columns dey give 256 Odù, and the count na exactly 256, no be more. The order of columns matter. The two casting methods dey equivalent in terms of information. The mark patterns for the sixteen columns complete and dem no repeat. The system dey use base 2 for the exact sense say the atomic unit get two states and the figures na strings on top that unit. Bascom own 2^n formulation na scholar description of the combinatorics, wey dem publish for 1969 1.

Wetin anything for above no establish. Say the Yorùbá do arithmetic for base 2, wey be different claim from having two-state notation. Nothing dey the sources wey show say dem add, multiply, or manipulate Odù as numbers. The decimal values wey dey the table na conversions wey this corpus do, and no evidence dey say any diviner dey treat Èjì Ogbè as fifteen.

Say the ordering na numerical sequence. E no be so: if you read am as binary, the ranking dey run 15, 0, 6, 9, 8, 1, 12, 3, 14, 7, 13, 11, 2, 4, 10, 5, wey no be any numerical order wey person fit recognize. The ranking na by seniority according to the tradition own account, wey be mythological and wey dem discuss for file 03.

Say single mark equal 1 and double equal 0 instead of the opposite. The tradition no assign any numeral to any of the marks. Both mappings dey appear for modern writing, and to choose one na modern convention. This corpus choose one and label am as choice.

Say any of this one make Ifá computer, or proto-computer. Na addressing scheme over fixed corpus, without any operations wey dem define on the addresses, no storage wey dem dey write to during use, and no computation wey dem perform on the values. To call am binary information system na something wey person fit defend. But to call am computation no correct, and file 07 break down the popular versions of this claim one by one.

The structure dey impressive on its own. Society wey no get writing build randomising address space of exactly 256 slots, with physical procedure wey dey reliably produce uniform two-state outputs, rejection rule to discard invalid draws, fixed orientation convention to make the address ordered pair, and naming scheme wey dey allow any practitioner name any address from the marks. That one na real achievement for information design and e no need inflating.

History and evolution

The binary structure of Ifá divination represent oral information technology wey exist before European contact for West Africa. For the earliest documented forms wey oral traditions and archaeological records preserve for Ilé-Ifẹ̀, the system establish sixteen primary figures (ojú odù) and their 256 paired permutations 1. As the Ọ̀yọ́ Empire expand across Yorùbáland during the seventeenth and eighteenth centuries, Ọ̀yọ́ standardize the regional administration of divination guilds and support babaláwo, wey spread uniform notation and ranking across expanding territorial networks 1.

The nineteenth-century Yorùbá civil wars and the collapse of the Ọ̀yọ́ capital scatter traditional state institutions, displace lineages, and disperse diviners go different parts of southwestern Nigeria 1. At the same time, Christian missionary contact for mid-nineteenth century bring aggressive campaigns against indigenous knowledge systems. Early missionary accounts, like the ones wey James Johnson write, systematically misunderstand the binary notation, blow the count of figures reach thousands, and present the practice as heathen superstition instead of structured mnemonic inquiry 1.

Under British colonial rule for late nineteenth and early twentieth century, colonial authorities push Ifá go one side under anti-witchcraft laws, even though the internal mathematical structure and casting techniques still remain well-kept inside hereditary secret societies 3. After Nigeria get independence for 1960, academic studies by scholars like William Bascom and Wande Abimbola properly document the 256 Odù and correct colonial mathematical mistakes, showing the strong binary logic of the eight-bit system [S1, S2].

Today, the 256 Odù system dey work across the whole world for West Africa and everywhere inside African diaspora, including Cuban Santería, Lucumí, and Brazilian Candomblé traditions 3. For 2008, UNESCO recognize the Ifá divination system as Masterpiece of the Oral and Intangible Heritage of Humanity, as dem acknowledge how e still strong both for traditional community governance and for transnational digital networks 3.