A Comparison of Indus Script and Mycenaean Linear B

Linear B tablet Ce 162 from Knossos: tall sign left of center is OVIS, "sheep,"
and the dotted, rayed circle represents 10,000 (Chadwick et al 1986: 79).
The other marks are numerals: 9,000 (rayed circles), 700 (circles), perhaps
90 (horizontal strokes), and 4 (verticals) = 9,794.

Linear B is a logo-syllabic script of Bronze Age Greece, deciphered in 1953 largely by Michael Ventris (Robinson2009: 95-103).  The basic syllabary portion contains 60 common signs, with an additional 27 occurring less frequently.  There are also four basic symbols used for writing numerals: a vertical stroke for each unit, a horizontal stroke for each ten, a circle for each hundred, and for each thousand a circle with a vertical stroke above and below as well as a horizontal stroke on either side.  In addition, there are a number of logograms, which tend to occur less frequently than the syllabic or numerical signs.  If one assumes about a hundred logograms, this would make a total of 191 signs.  Thus, the overall sign list is considerably shorter than what one finds in Mesopotamian cuneiform, in Egyptian hieroglyphs, or in Indus script, each of which has several hundred symbols.  But an examination of Linear B may still prove informative for the student of Indus symbols, if only as a contrast.

The following table indicates the syllabic signs that appear in Linear B (all either vowels or combinations of consonant plus vowel), with the frequency of each one.  The numbers in the table represent my count of occurrences in the first 100 pages of Corpus of Mycenaean Inscriptions from Knossos (Chadwick et al 1986, Vol. I).  The most frequent sign represents the vowel a, occurring 69 times or 4.6% of the total.  The least frequent syllables (of those that actually occur in this sample) appear less than 1% of the time.  This is not a large frequency range.  Compare the frequencies of signs in the Indus script, where most symbols appear fewer than 10 times and comprise less than 1% of the whole, versus a small core of commonly occurring symbols that, together, comprise nearly half the whole.

Vowel/ Consonant	a	e	i	o	u
(pure vowels)	69 (4.6%)	36 (2.4%)	40 (2.7%)	46 (3.1%)	32 (2.1%)
d	42 (2.8%)	22 (1.5%)	17 (1.1%)	10 (0.7%)	3 (0.2%)
j	42 (2.8%)	9 (0.6%)	--	62 (4.1%)	0
k	45 (3.0%)	25 (1.7%)	16 (1.1%)	41 (2.7%)	20 (4.0%)
m	14 (0.9%)	31 (2.1%)	14 (0.9%)	23 (1.5%)	0
n	19 (1.3%)	21 (1.4%)	30 (2.0%)	36 (2.4%)	9 (0.6%)
p	48 (3.2%)	19 (1.3%)	15 (1.0%)	13 (0.9%)	5 (0.3%)
q	8 (0.5%)	18 (1.2%)	5 (0.3%)	12 (0.1%)	--
r	44 (2.9%)	41 (2.7%)	48 (3.2%)	38 (2.5%)	16 (1.1%)
s	13 (0.9%)	6 (0.4%)	35 (2.3%)	22 (1.5%)	4 (0.3%)
t	62 (4.1%)	32 (2.1%)	22 (1.5%)	53 (3.5%)	8 (0.5%)
w	33 (2.2%)	28 (1.9%)	15 (1.0%)	43 (2.9%)	--
z	4 (0.3%)	10 (0.7%)	--	2 (0.1%)	--

Table 1.  Frequency of syllabic signs in a sample of Linear B texts from Knossos.

 

In this same sample, logographs also occur, but they are generally less frequent than the syllabic signs.  Logographs for people and animals are fairly common, as are commodities associated with them:

Sign	Meaning	Frequency
BOS	ox, cow, bull	18 (8.8%)
CAP	goat	6 (2.9%)
EQU	horse	11 (5.4%)
OVIS	sheep	27 (13.5%)
SUS	pig	3 (1.5%)
VIR	man	19 (9.3%)
MUL	woman	5 (2.5%)
BIG	chariot	9 (4.4%)
LANA	wool	1 (0.5%)

Table 2.  Frequency of some logographs in sample of Linear B texts from Knossos.

Other commodities include domesticated plants, four different types of vessels containing other products (each occurring once in this sample), and the occasional manufactured object.

Sign	Meaning	Frequency
AES	bronze	1 (0.5%)
CROC	saffron	1 (0.5%)
CYP	copper	3 (1.5%)
FAR	flour/meal	13 (6.4%)
GRA	grain/seed	7 (3.4%)
HORD	barley	21 (10.3%)
OLE	oil	28 (13.7%)
OLIV	olives	2 (1.0%)
TELA	cloth	3 (1.5%)
TUN	tunic (armor)	12 (5.9%)
VIN	wine	14 (6.8%)

Table 3.  Frequency of additional logographs in a sample of Linear B texts from Knossos.

Percentages shown only tell part of the story, though, since I have separated syllabic signs from logographs.  If I combine them, the percentages all decrease, with the most frequent sign comprising 4.1% of the total and the least frequent about 0.1%.  In this sample, there are three singletons, all of them logographs.  There are actually three other symbols not shown in these tables, which are also singletons or quite rare.  Their meaning remains uncertain, so they are transliterated with numerical designators, shown in italics, with an asterisk preceding (nine such singletons).  Some could be logographs.  But at least one is probably a syllabic sign because it apparently occurs as part of a word, between two other syllabic signs (designated *22 with 6 occurrences).  Adding these in with the others reduces the percentages of frequency slightly.

Linear B tablet Dl(1) 947 + 7626: e-se-re-e-jo OVIS 80 ki OVIS 10  LANA 11
si-da-jo / ma-ri  o ki OVIS 70   o LANA 5 (recording numbers of sheep and
amounts of wool) (Chadwick et al 1986: 946).  Note the "11," on far right, top row,
a number which never appears in Indus inscriptions.

As noted, syllabic signs tend to occur more often than logographs, but there are plenty of exceptions.  Two of the most common logographs, OLE and HORD,  each occur more than 20 times, so that each comprises more than 10% of the whole sample.  And while the rare signs are more likely to be logographs, this is not always the case, either.  There are also a few rare syllabic signs, especially those ending with –u (du, nu, pu, su, and tu each appear fewer than 10 times and neither ju nor mu shows up at all in this sample).  With this amount of overlap in frequency between the two categories of signs, clearly the frequency of a sign alone cannot tell you whether it is a phonetic or a logographic symbol. 

The two categories of signs are not entirely separate, either, since several syllabic signs also function as logograms.  For example, the syllable ni commonly functions as a logograph for figs.  Context distinguishes the two categories in Linear B.  Logographs occur most often next to numerals.  Syllabic signs, on the other hand, generally occur in small groups, as they spell out words.

Linear B tablet fragment V(2) 159 + fr. (Chadwick et al 1986: 159) with
four rows of text: 83 ru-ki-to 157 (top complete row); 55 ti-ri-to 45 (2nd
row); -so 34 pu2-te-re / wa-si-ro 18 (3rd row); 10 ra-pte-re 20 (bottom row).
Note the distinct appearance of tens (horizontal strokes) & ones (verticals).

A similar pattern has been noted before in connection with Egyptian hieroglyphs: the most frequent symbols tend to be those providing phonetic information, with the logographs occurring less often.  But a few specific logographs are actually quite common (e.g., the sign ntr, “god” in the Book of the Dead).  And some of the phonetic symbols are rare (especially certain triliterals).

The study of one writing system is not necessarily informative concerning details of an unrelated writing system (especially true of proto-writing).  Still, such an examination can induce caution in studying Indus symbols, especially for those who see this as a logo-syllabic script (the position of a number of researchers).  Specifically, one can determine which Indus signs are most common and which are rare, simply by counting.  But having done this, one cannot automatically assume that common signs indicate phonetic information (whether syllabic or otherwise).  Nor can one automatically assume that rare symbols are always logographs.  In fact, the two categories of signs – phonetic symbols versus logographs – overlap in Linear B and in Egyptian hieroglyphs, since a given symbol may have both functions.  This characteristic – where one sign has more than one function – is also found in cuneiform.

Beyond these general observations, note that the frequency data from Linear B is quite different from that of Indus signs.  There are some singletons in my sample of Linear B, but these form a small proportion of the total.  And among the frequently occurring signs, there is no individual symbol that appears much more often than all the others.  In Indus script, in contrast, the proportion of singletons is quite large – well over half of the signs are either singletons or quite rare.  The proportion of frequently occurring Indus signs is quite small and one sign – the POT – occurs considerably more often than any other sign.  These differences in the patterns of frequency provide a strong indication that Indus signs do not function as a logo-syllabic writing system.

The frequency of Linear B numerals is also interesting:

Numeral	Occurrences	Percent
ONE	693	42.9
TWO	261	16.2
THREE	119	7.4
FOUR	93	5.8
FIVE	87	5.4
SIX	71	4.4
SEVEN	43	2.7
EIGHT	35	2.2
NINE	30	1.9
TEN	144	8.9
ELEVEN	10	0.1
TWELVE	29	1.8
TOTAL	1,615	100%*

Table 4.  Frequency of numerals in Linear B texts from Knossos

(*total percentages actually add up to 99.7% due to rounding).

The pattern shown here is clearly that demonstrated by Dehaene and Mehler’s (1992: 1-29) study of number frequency in modern languages: “In all languages, the frequency with which number words are printed or uttered decreases with magnitude, aside from local increases for the round numbers 10, 12, 15, 20, 50, and 100.”  In other words, the smaller the number the higher its frequency and, conversely, the larger the number the lower its frequency.  Note that in this selection of Linear B tablets, the local increases cited by Dehaene and Mehler for 15, 20, and 50 do not appear (numerical data on Linear B includes all texts in volume I).

Had I included ten and its multiples in this table, it would have shown a similar pattern, with ten showing the highest frequency (144 occurrences) and ninety the lowest (11 occurrences).  Similarly, a table of the hundreds would show 100 with the highest frequency (81 occurrences) versus the very low 900 (4 occurrences).  In the case of the Linear B tablets from Knossos, though, 800 does not occur and 600 only appears once, no doubt due to chance.  There are also 3 occurrences of the number 1,000 and 2 of 10,000, but other thousands are essentially lacking.

Indus seal B-10 with an apparent numeral at the top, SEVEN QUOTES,
and perhaps an enumerated item before the iconic animal, TRI-FORK.

Now, the Linear B tablets are basically accounting documents, so the appearance of more than one number on a single tablet is quite common.  In fact, this is the case on nearly all the complete tablets.  The numerals indicate quantities, preceded by logographs indicating the commodities, which are preceded in turn by phonetic signs.  The phonetically spelled words are sometimes recognizable as place names, at other times clearly the names of persons, at still other times the names of deities.  There are also some symbols, sometimes occurring in between the logographs and the numerals, that represent measures.  For example, if we were to write the English phrase “10 gallons of milk” in the Linear B manner, we would begin with a symbol representing milk (let us say “M”), follow this with a sign for the gallon measure (“g”), and conclude with a symbol for the numeral ten (“X”):  M g X.

A document from Knossos in Linear B concerning cloth is KN Ld 571 (Hooker 1980: 96).  In large signs on the left is the three-syllable word pa-we-a (resembling a shish kebab, backward “S,” and capital “T” with serifs).  This represents an archaic Greek word pharwea (φαρFεα) meaning “pieces of cloth.”  To the right are two rows of smaller symbols.  The upper row is transliterated pe-ne-we-ta a-ro₂-a *158 1; the lower row e-qe-si-ja re-u-ko-nu-ka TELA 25.  The first word, pe-ne-we-ta, remains uninterpreted, but a-ro₂-a is the Mycenaean version of areion (αρειων), “better, of superior quality.”  The next symbol is enigmatic.  Since its definition has not been established, it is transcribed with a numerical designator with an asterisk before it.  It may indicate a standard container of some kind (1980: 97).  The “1” which follows is a numeral indicating the quantity of such containers, namely one.  In the lower row, e-qe-si-ja that shows the following object is intended for a person entitled the epetas (επετας).  The word after this is re-u-ko-nu-ka (λευκονυχα) which means “(decorated with) white ονυχης,” probably decorated with a white pattern resembling claws or hooves.  Capital letters indicate a logograph, TELA, meaning “cloth,” followed by the numeral 25.  Thus, there are 25 cloths for the epetas with a decorative pattern of hooves or claws, as well as one container of superior quality, perhaps for someone entitled pe-ne-we-ta.  This is the typical format of Linear B documents – several syllabic signs, a logograph, and a numeral in each line.  A single clay tablet may contain only one line of text like this, or many such lines.

Can a study of the Linear B texts provide useful information for interpreting Indus inscriptions?  I doubt it.  The basic format is quite different.  Indus seals lack apparent numerals more often than not.  When “numerals” do occur, they may be at the beginning of an inscription, in the middle, or at the end.  Usually when a “numeral” appears, there is only one.  Even when there seem to be two or more “numerals,” they do not line up the way numerals do in Linear B.  That is, there is no standard format in the Indus inscriptions comparable to this one just described for Linear B.

REFERENCES

Chadwick, J., L. Godart, J.T. Killen, J.-P. Olivier, A. Sacconi, I.A. Sakellarakis. 1986. Corpus of Mycenaean Inscriptions from Knossos. Vol. I (1-1063).  Cambridge: Cambridge University Press.

Dehaene, S. and J. Mehler. 1992. "Cross-linguistic regularities in the frequency of number words," in Cognition (43): 1-29.

Hooker, J.T. 1980. Linear B: An Introduction. London: Bristol Classical Press.