<s>
There	O
have	O
been	O
numerous	O
attempts	O
to	O
decipher	O
the	O
rongorongo	B-Language
script	O
of	O
Easter	O
Island	O
since	O
its	O
discovery	O
in	O
the	O
late	O
nineteenth	O
century	O
.	O
</s>
<s>
As	O
with	O
most	O
undeciphered	B-Algorithm
scripts	I-Algorithm
,	O
many	O
of	O
the	O
proposals	O
have	O
been	O
fanciful	O
.	O
</s>
<s>
The	O
evidence	O
is	O
weak	O
that	O
rongorongo	B-Language
directly	O
represents	O
the	O
Rapa	O
Nui	O
language	O
–	O
that	O
is	O
,	O
that	O
it	O
is	O
a	O
true	O
writing	O
system	O
–	O
and	O
oral	O
accounts	O
report	O
that	O
experts	O
in	O
one	O
category	O
of	O
tablet	O
were	O
unable	O
to	O
read	O
other	O
tablets	O
,	O
suggesting	O
either	O
that	O
rongorongo	B-Language
is	O
not	O
a	O
unified	O
system	O
,	O
or	O
that	O
it	O
is	O
proto-writing	O
that	O
requires	O
the	O
reader	O
to	O
already	O
know	O
the	O
text	O
.	O
</s>
<s>
Assuming	O
that	O
rongorongo	B-Language
is	O
writing	O
,	O
there	O
are	O
three	O
serious	O
obstacles	O
to	O
decipherment	O
:	O
the	O
small	O
number	O
of	O
remaining	O
texts	O
,	O
comprising	O
only	O
15,000	O
legible	O
glyphs	B-Application
;	O
the	O
lack	O
of	O
context	O
in	O
which	O
to	O
interpret	O
the	O
texts	O
,	O
such	O
as	O
illustrations	O
or	O
parallel	O
texts	O
which	O
can	O
be	O
read	O
;	O
and	O
the	O
fact	O
that	O
the	O
modern	O
Rapa	O
Nui	O
language	O
is	O
heavily	O
mixed	O
with	O
Tahitian	O
and	O
is	O
unlikely	O
to	O
closely	O
reflect	O
the	O
language	O
of	O
the	O
tablets	O
—	O
especially	O
if	O
they	O
record	O
a	O
specialized	O
register	O
such	O
as	O
incantations	O
—	O
while	O
the	O
few	O
remaining	O
examples	O
of	O
the	O
old	O
language	O
are	O
heavily	O
restricted	O
in	O
genre	O
and	O
may	O
not	O
correspond	O
well	O
to	O
the	O
tablets	O
either	O
.	O
</s>
<s>
Since	O
a	O
proposal	O
by	O
Butinov	O
and	O
Knorozov	O
in	O
the	O
1950s	O
,	O
the	O
majority	O
of	O
philologists	O
,	O
linguists	O
and	O
cultural	O
historians	O
have	O
taken	O
the	O
line	O
that	O
rongorongo	B-Language
was	O
not	O
true	O
writing	O
but	O
proto-writing	O
,	O
that	O
is	O
,	O
an	O
ideographic	O
-	O
and	O
rebus-based	O
mnemonic	O
device	O
,	O
such	O
as	O
the	O
Dongba	O
script	O
of	O
the	O
Nakhi	O
people	O
,	O
which	O
would	O
in	O
all	O
likelihood	O
make	O
it	O
impossible	O
to	O
decipher	O
.	O
</s>
<s>
Of	O
those	O
who	O
have	O
attempted	O
to	O
decipher	O
rongorongo	B-Language
as	O
a	O
true	O
writing	O
system	O
,	O
the	O
vast	O
majority	O
have	O
assumed	O
it	O
was	O
logographic	O
,	O
a	O
few	O
that	O
it	O
was	O
syllabic	O
or	O
mixed	O
.	O
</s>
<s>
In	O
1868	O
the	O
Bishop	O
of	O
Tahiti	O
,	O
Florentin-Étienne	O
Jaussen	O
,	O
received	O
a	O
gift	O
from	O
recent	O
converts	O
on	O
Easter	O
Island	O
:	O
a	O
long	O
cord	O
of	O
human	O
hair	O
wound	O
around	O
a	O
discarded	O
rongorongo	B-Language
tablet	O
.	O
</s>
<s>
William	O
J	O
.	O
Thomson	O
,	O
paymaster	O
on	O
the	O
USS	O
Mohican	O
,	O
spent	O
twelve	O
days	O
on	O
Easter	O
Island	O
from	O
19	O
December	O
to	O
30	O
December	O
1886	O
,	O
during	O
which	O
time	O
he	O
made	O
an	O
impressive	O
number	O
of	O
observations	O
,	O
including	O
some	O
which	O
are	O
of	O
interest	O
for	O
the	O
decipherment	O
of	O
the	O
rongorongo	B-Language
.	O
</s>
<s>
He	O
had	O
been	O
the	O
steward	O
of	O
King	O
Ngaꞌara	O
,	O
the	O
last	O
king	O
said	O
to	O
have	O
had	O
knowledge	O
of	O
writing	O
,	O
and	O
although	O
he	O
was	O
not	O
able	O
to	O
write	O
himself	O
,	O
he	O
knew	O
many	O
of	O
the	O
rongorongo	B-Language
chants	O
and	O
was	O
able	O
to	O
read	O
at	O
least	O
one	O
memorized	O
text	O
.	O
</s>
<s>
|	O
R	B-Algorithm
(	O
Small	O
Washington	O
)	O
?	O
</s>
<s>
The	O
verses	O
of	O
Atua	O
Matariri	O
are	O
of	O
the	O
form	O
X	O
ki	O
ꞌai	O
ki	O
roto	O
Y	O
,	O
ka	O
pû	O
te	O
Z	O
"	O
X	O
,	O
by	O
mounting	O
into	O
Y	O
,	O
let	O
Z	O
come	O
forth	O
"	O
,	O
</s>
<s>
For	O
example	O
,	O
the	O
composition	O
of	O
the	O
Chinese	O
character	O
銅	O
tóng	O
"	O
copper	O
"	O
may	O
be	O
described	O
as	O
"	O
add	O
同	O
tóng	O
to	O
金	O
jīn	O
to	O
make	O
銅	O
tóng	O
"	O
(	O
meaning	O
"	O
add	O
Together	O
to	O
Metal	O
to	O
make	O
Copper	O
"	O
)	O
,	O
which	O
is	O
also	O
nonsense	O
when	O
taken	O
literally	O
.	O
</s>
<s>
He	O
hypothesizes	O
that	O
the	O
Atua	O
Matariri	O
chant	O
which	O
Ure	O
had	O
heard	O
in	O
his	O
youth	O
,	O
although	O
unconnected	O
to	O
the	O
particular	O
tablet	O
for	O
which	O
he	O
recited	O
it	O
,	O
was	O
a	O
genuine	O
rongorongo	B-Language
chant	O
:	O
A	O
mnemonic	O
which	O
taught	O
students	O
how	O
the	O
glyphs	B-Application
were	O
composed	O
.	O
</s>
<s>
Perhaps	O
due	O
to	O
the	O
cost	O
of	O
casting	O
special	O
type	B-Protocol
for	O
rongorongo	B-Language
,	O
no	O
method	O
,	O
analysis	O
,	O
or	O
sound	O
values	O
of	O
the	O
individual	O
glyphs	B-Application
were	O
ever	O
published	O
.	O
</s>
<s>
In	O
1932	O
the	O
Hungarian	O
Vilmos	O
Hevesy	O
(	O
Guillaume	O
de	O
Hevesy	O
)	O
published	O
an	O
article	O
claiming	O
a	O
relationship	O
between	O
rongorongo	B-Language
and	O
the	O
Indus	O
Valley	O
script	O
,	O
based	O
on	O
superficial	O
similarities	O
of	O
form	O
.	O
</s>
<s>
Due	O
to	O
the	O
lack	O
of	O
an	O
accessible	O
rongorongo	B-Language
corpus	O
for	O
comparison	O
,	O
it	O
was	O
not	O
apparent	O
that	O
several	O
of	O
the	O
rongorongo	B-Language
glyphs	B-Application
illustrated	O
in	O
Hevesy	O
's	O
publications	O
were	O
spurious	O
.	O
</s>
<s>
At	O
least	O
a	O
score	O
of	O
decipherments	O
have	O
been	O
claimed	O
since	O
then	O
,	O
none	O
of	O
which	O
have	O
been	O
accepted	O
by	O
other	O
rongorongo	B-Language
epigraphers	O
.	O
</s>
<s>
More	O
rigorous	O
than	O
most	O
attempts	O
,	O
she	O
restricts	O
each	O
glyph	B-Application
to	O
a	O
single	O
logographic	O
reading	O
.	O
</s>
<s>
For	O
example	O
,	O
tablet	O
P	B-Algorithm
begins	O
(	O
with	O
each	O
rongorongo	B-Language
ligature	O
set	O
off	O
with	O
a	O
comma	O
in	O
the	O
translation	O
)	O
:	O
</s>
<s>
(	O
Pozdniakov	O
has	O
demonstrated	O
that	O
these	O
are	O
graphic	O
variants	O
of	O
the	O
same	O
glyph	B-Application
.	O
)	O
</s>
<s>
As	O
it	O
was	O
,	O
Fedorova	O
's	O
catalog	O
consisted	O
of	O
only	O
130	O
glyphs	B-Application
;	O
Pozdniakov	O
's	O
additional	O
allography	O
would	O
have	O
reduced	O
that	O
number	O
and	O
made	O
her	O
interpretation	O
even	O
more	O
repetitive	O
.	O
</s>
<s>
Such	O
extreme	O
repetition	O
is	O
a	O
problem	O
with	O
all	O
attempts	O
to	O
read	O
rongorongo	B-Language
as	O
a	O
logographic	O
script	O
.	O
</s>
<s>
Many	O
recent	O
scholars	O
are	O
of	O
the	O
opinion	O
that	O
,	O
while	O
many	O
researchers	O
have	O
made	O
modest	O
incremental	O
contributions	O
to	O
the	O
understanding	O
of	O
rongorongo	B-Language
,	O
notably	O
Kudrjavtsev	O
et	O
al.	O
,	O
Butinov	O
and	O
Knorozov	O
,	O
and	O
Thomas	O
Barthel	O
,	O
the	O
attempts	O
at	O
actual	O
decipherment	O
,	O
such	O
as	O
those	O
of	O
Fedorova	O
here	O
or	O
of	O
Fischer	O
below	O
,	O
"	O
are	O
not	O
accompanied	O
by	O
the	O
least	O
justification	O
"	O
.	O
</s>
<s>
James	O
Park	O
Harrison	O
,	O
a	O
council	O
member	O
of	O
the	O
Anthropological	O
Institute	O
of	O
Great	O
Britain	O
and	O
Ireland	O
,	O
noticed	O
that	O
lines	O
Gr3	O
–	O
7	O
of	O
the	O
Small	B-Algorithm
Santiago	I-Algorithm
tablet	I-Algorithm
featured	O
a	O
compound	O
glyph	B-Application
,	O
(	O
a	O
sitting	O
figure	O
holding	O
a	O
rod	O
with	O
a	O
line	O
of	O
chevrons	O
(	O
a	O
garland	O
?	O
)	O
</s>
<s>
repeated	O
31	O
times	O
,	O
each	O
time	O
followed	O
by	O
one	O
to	O
half	O
a	O
dozen	O
glyphs	B-Application
before	O
its	O
next	O
occurrence	O
.	O
</s>
<s>
Barthel	O
later	O
found	O
this	O
pattern	O
on	O
tablet	O
K	B-Algorithm
,	O
which	O
is	O
a	O
paraphrase	O
of	O
Gr	O
(	O
in	O
many	O
of	O
the	O
K	B-Algorithm
sequences	O
the	O
compound	O
is	O
reduced	O
to	O
as	O
well	O
as	O
on	O
A	O
,	O
where	O
it	O
sometimes	O
appears	O
as	O
380.1.3	O
and	O
sometimes	O
as	O
380.1	O
;	O
on	O
C	O
,	O
E	O
,	O
and	O
S	O
as	O
380.1	O
;	O
and	O
,	O
with	O
the	O
variant	O
on	O
N	B-Algorithm
.	O
In	O
places	O
it	O
appears	O
abbreviated	O
as	O
or	O
without	O
the	O
human	O
figure	O
,	O
but	O
parallels	O
in	O
the	O
texts	O
suggest	O
these	O
have	O
the	O
same	O
separating	O
function	O
.	O
</s>
<s>
The	O
group	O
later	O
noticed	O
that	O
tablet	O
K	B-Algorithm
was	O
a	O
close	O
paraphrase	O
of	O
the	O
recto	O
of	O
G	B-Algorithm
.	O
Kudrjavtsev	O
wrote	O
up	O
their	O
findings	O
,	O
which	O
were	O
published	O
posthumously	O
.	O
</s>
<s>
Numerous	O
other	O
parallel	O
,	O
though	O
shorter	O
,	O
sequences	O
have	O
since	O
been	O
identified	O
through	O
statistical	O
analysis	O
,	O
with	O
texts	O
N	B-Algorithm
and	O
R	B-Algorithm
found	O
to	O
be	O
composed	O
almost	O
entirely	O
of	O
phrases	O
shared	O
with	O
other	O
tablets	O
,	O
though	O
not	O
in	O
the	O
same	O
order	O
.	O
</s>
<s>
Ligatures	O
:	O
Parallel	O
texts	O
Pr4	O
–	O
5	O
(	O
top	O
)	O
and	O
Hr5	O
(	O
bottom	O
)	O
show	O
that	O
a	O
figure	O
(	O
glyph	B-Application
holding	O
an	O
object	O
and	O
in	O
P	B-Algorithm
may	O
be	O
fused	O
into	O
a	O
ligature	O
in	O
H	B-Algorithm
,	O
where	O
the	O
object	O
replaces	O
either	O
the	O
figure	O
's	O
head	O
or	O
its	O
hand	O
.	O
</s>
<s>
(	O
Elsewhere	O
in	O
these	O
texts	O
,	O
animal	O
figures	O
are	O
reduced	O
to	O
a	O
distinctive	O
feature	O
such	O
as	O
a	O
head	O
or	O
arm	O
when	O
they	O
fuse	O
with	O
a	O
preceding	O
glyph	B-Application
.	O
)	O
</s>
<s>
Here	O
also	O
are	O
the	O
two	O
hand	O
shapes	O
(	O
glyphs	B-Application
and	O
which	O
would	O
later	O
be	O
established	O
as	O
allographs	O
.	O
</s>
<s>
Three	O
of	O
the	O
four	O
human	O
and	O
turtle	O
figures	O
at	O
left	O
have	O
arm	O
ligatures	O
with	O
an	O
orb	O
(	O
glyph	B-Application
which	O
Pozdniakov	O
found	O
often	O
marks	O
a	O
phrase	O
boundary	O
.	O
</s>
<s>
Now	O
,	O
if	O
the	O
repeated	O
independent	O
glyph	B-Application
is	O
a	O
title	O
,	O
such	O
as	O
"	O
king	O
"	O
,	O
and	O
if	O
the	O
repeated	O
attached	O
glyph	B-Application
is	O
a	O
patronymic	O
marker	O
,	O
then	O
this	O
means	O
something	O
like	O
:	O
</s>
<s>
If	O
it	O
is	O
correct	O
,	O
then	O
,	O
first	O
,	O
we	O
can	O
identify	O
other	O
glyph	B-Application
sequences	O
which	O
constitute	O
personal	O
names	O
.	O
</s>
<s>
Second	O
,	O
the	O
Santiago	B-Algorithm
Staff	I-Algorithm
would	O
consist	O
mostly	O
of	O
persons	O
 '	O
names	O
as	O
it	O
bears	O
564	O
occurrences	O
of	O
glyph	B-Application
76	O
,	O
the	O
putative	O
patronymic	O
marker	O
,	O
one	O
fourth	O
of	O
the	O
total	O
of	O
2320	O
glyphs	B-Application
.	O
</s>
<s>
The	O
Santiago	B-Algorithm
Staff	I-Algorithm
,	O
with	O
63	O
occurrences	O
of	O
glyph	B-Application
a	O
rebus	O
for	O
îka	O
"	O
victim	O
"	O
,	O
would	O
then	O
be	O
in	O
part	O
a	O
kohau	O
îka	O
(	O
list	O
of	O
war	O
casualties	O
)	O
.	O
</s>
<s>
German	O
ethnologist	O
Thomas	O
Barthel	O
,	O
who	O
first	O
published	O
the	O
rongorongo	B-Language
corpus	O
,	O
identified	O
three	O
lines	O
on	O
the	O
recto	O
(	O
side	O
a	O
)	O
of	O
tablet	O
C	O
,	O
also	O
known	O
as	O
Mamari	O
,	O
as	O
a	O
lunar	O
calendar	O
.	O
</s>
<s>
In	O
1995	O
independent	O
linguist	O
Steven	O
Fischer	O
,	O
who	O
also	O
claims	O
to	O
have	O
deciphered	O
the	O
enigmatic	O
Phaistos	O
Disc	O
,	O
announced	O
that	O
he	O
had	O
cracked	O
the	O
rongorongo	B-Language
"	O
code	O
"	O
,	O
making	O
him	O
the	O
only	O
person	O
in	O
history	O
to	O
have	O
deciphered	O
two	O
such	O
scripts	O
.	O
</s>
<s>
Fischer	O
identified	O
glyph	B-Application
76	O
as	O
a	O
phallus	O
and	O
the	O
text	O
of	O
the	O
Santiago	B-Algorithm
Staff	I-Algorithm
as	O
a	O
creation	O
chant	O
consisting	O
of	O
hundreds	O
of	O
repetitions	O
of	O
X	O
–	O
phallus	O
Y	O
Z	O
,	O
which	O
he	O
interpreted	O
as	O
X	O
copulated	O
with	O
Y	O
,	O
there	O
issued	O
forth	O
Z	O
.	O
</s>
<s>
about	O
half-way	O
through	O
line	O
12	O
of	O
the	O
Santiago	B-Algorithm
Staff	I-Algorithm
.	O
</s>
<s>
Fischer	O
interpreted	O
glyph	B-Application
606	O
as	O
"	O
bird	O
"	O
+	O
"	O
hand	O
"	O
,	O
with	O
the	O
phallus	O
attached	O
as	O
usual	O
at	O
its	O
lower	O
right	O
;	O
glyph	B-Application
700	O
as	O
"	O
fish	O
"	O
;	O
and	O
glyph	B-Application
8	O
as	O
"	O
sun	O
"	O
.	O
</s>
<s>
This	O
recitation	O
is	O
a	O
litany	O
where	O
each	O
verse	O
has	O
the	O
form	O
X	O
,	O
ki	O
ꞌai	O
ki	O
roto	O
ki	O
Y	O
,	O
ka	O
pû	O
te	O
Z	O
,	O
literally	O
"	O
X	O
having	O
been	O
inside	O
Y	O
the	O
Z	O
comes	O
forward	O
"	O
.	O
</s>
<s>
Fischer	O
proposed	O
that	O
the	O
glyph	B-Application
sequence	O
606.76	O
700	O
8	O
,	O
literally	O
MANU:MAꞋU.ꞋAI	O
ÎKA	O
RAꞋÂ	O
"	O
bird:hand.penis	O
fish	O
sun	O
"	O
,	O
had	O
the	O
analogous	O
phonetic	O
reading	O
of	O
:	O
</s>
<s>
However	O
,	O
in	O
the	O
majority	O
of	O
texts	O
glyph	B-Application
76	O
is	O
not	O
common	O
,	O
and	O
Fischer	O
proposed	O
that	O
these	O
were	O
a	O
later	O
,	O
more	O
developed	O
stage	O
of	O
the	O
script	O
,	O
where	O
the	O
creation	O
chants	O
had	O
been	O
abbreviated	O
to	O
X	O
Y	O
Z	O
and	O
omit	O
the	O
phallus	O
.	O
</s>
<s>
He	O
concluded	O
that	O
85%	O
of	O
the	O
rongorongo	B-Language
corpus	O
consisted	O
of	O
such	O
creation	O
chants	O
,	O
and	O
that	O
it	O
was	O
only	O
a	O
matter	O
of	O
time	O
before	O
rongorongo	B-Language
would	O
be	O
fully	O
deciphered	O
.	O
</s>
<s>
When	O
Andrew	O
Robinson	O
checked	O
the	O
claimed	O
pattern	O
,	O
he	O
found	O
that	O
"	O
Close	O
inspection	O
of	O
the	O
Santiago	B-Algorithm
Staff	I-Algorithm
reveals	O
that	O
only	O
63	O
out	O
of	O
the	O
113	O
 [ sic ] 	O
sequences	O
on	O
the	O
staff	O
fully	O
obey	O
the	O
triad	O
structure	O
(	O
and	O
63	O
is	O
the	O
maximum	O
figure	O
,	O
giving	O
every	O
Fischer	O
attribution	O
the	O
benefit	O
of	O
the	O
doubt	O
)	O
.	O
"	O
</s>
<s>
Glyph	B-Application
occurs	O
sometimes	O
in	O
isolation	O
,	O
sometimes	O
compounded	O
with	O
itself	O
,	O
and	O
sometimes	O
in	O
the	O
'	O
wrong	O
 '	O
part	O
(	O
or	O
even	O
all	O
parts	O
)	O
of	O
the	O
triplets	O
.	O
</s>
<s>
Pozdniakov	O
and	O
Pozdniakov	O
calculated	O
that	O
altogether	O
the	O
four	O
glyphs	B-Application
of	O
Fischer	O
's	O
primary	O
example	O
make	O
up	O
20%	O
of	O
the	O
corpus	O
.	O
</s>
<s>
Computational	O
linguist	O
Richard	O
Sproat	O
could	O
not	O
replicate	O
the	O
parallels	O
Fischer	O
claimed	O
between	O
the	O
Santiago	B-Algorithm
Staff	I-Algorithm
and	O
the	O
other	O
texts	O
.	O
</s>
<s>
He	O
automated	O
the	O
search	O
for	O
string	B-Algorithm
matches	I-Algorithm
between	O
the	O
texts	O
and	O
found	O
that	O
the	O
staff	O
stood	O
alone	O
:	O
</s>
<s>
In	O
the	O
1950s	O
,	O
Butinov	O
and	O
Knorozov	O
had	O
performed	O
a	O
statistical	O
analysis	O
of	O
several	O
rongorongo	B-Language
texts	O
and	O
had	O
concluded	O
that	O
either	O
the	O
language	O
of	O
the	O
texts	O
was	O
not	O
Polynesian	O
,	O
or	O
that	O
it	O
was	O
written	O
in	O
a	O
condensed	O
telegraphic	O
style	O
,	O
because	O
it	O
contained	O
no	O
glyphs	B-Application
comparable	O
in	O
frequency	O
to	O
Polynesian	O
grammatical	O
particles	O
such	O
as	O
the	O
Rapanui	O
articles	O
te	O
and	O
he	O
or	O
the	O
preposition	O
ki	O
.	O
</s>
<s>
These	O
findings	O
have	O
since	O
been	O
used	O
to	O
argue	O
that	O
rongorongo	B-Language
is	O
not	O
a	O
writing	O
system	O
at	O
all	O
,	O
but	O
mnemonic	O
proto-writing	O
.	O
</s>
<s>
The	O
problem	O
,	O
as	O
Butinov	O
and	O
Knorozov	O
,	O
and	O
Barthel	O
himself	O
,	O
had	O
admitted	O
,	O
was	O
that	O
in	O
many	O
cases	O
distinct	O
numerical	O
codes	O
had	O
been	O
assigned	O
to	O
ligatures	O
and	O
allographs	O
,	O
as	O
if	O
these	O
were	O
independent	O
glyphs	B-Application
.	O
</s>
<s>
In	O
2011	O
,	O
Pozdniakov	O
released	O
a	O
pre-press	O
publication	O
analyzing	O
Text	O
E	O
Keiti	O
,	O
including	O
a	O
glyph-by-glyph	O
comparison	O
of	O
the	O
transcription	O
in	O
Barthel	O
(	O
1958	O
)	O
,	O
with	O
misidentified	O
glyphs	B-Application
corrected	O
per	O
Horley	O
(	O
2010	O
)	O
.	O
</s>
<s>
He	O
observed	O
that	O
all	O
these	O
texts	O
but	O
I	O
and	O
G	B-Algorithm
verso	O
consist	O
predominantly	O
of	O
shared	O
phrases	O
(	O
sequences	O
of	O
glyphs	B-Application
)	O
,	O
which	O
occur	O
in	O
different	O
orders	O
and	O
contexts	O
on	O
different	O
tablets	O
.	O
</s>
<s>
Phrasing	O
:	O
Variants	O
of	O
this	O
twenty-glyph	O
phrase	O
,	O
all	O
missing	O
some	O
of	O
these	O
glyphs	B-Application
or	O
adding	O
others	O
,	O
are	O
found	O
twelve	O
times	O
,	O
in	O
eight	O
of	O
the	O
thirteen	O
texts	O
Pozdniakov	O
tabulated	O
:	O
lines	O
Ab4	O
,	O
Cr2	O
–	O
3	O
,	O
Cv2	O
,	O
Cv12	O
,	O
Ev3	O
,	O
Ev6	O
,	O
Gr2	O
–	O
3	O
,	O
Hv12	O
,	O
Kr3	O
,	O
Ra6	O
,	O
Rb6	O
,	O
and	O
Sa1	O
.	O
</s>
<s>
These	O
shared	O
sequences	O
begin	O
and	O
end	O
with	O
a	O
notably	O
restricted	O
set	O
of	O
glyphs	B-Application
.	O
</s>
<s>
Contrasting	O
these	O
phrases	O
allowed	O
Pozdniakov	O
to	O
determine	O
that	O
some	O
glyphs	B-Application
occur	O
in	O
apparent	O
free	O
variation	O
both	O
in	O
isolation	O
and	O
as	O
components	O
of	O
ligatures	O
.	O
</s>
<s>
Thus	O
he	O
proposed	O
that	O
the	O
two	O
hand	O
shapes	O
,	O
(	O
three	O
fingers	O
and	O
a	O
thumb	O
)	O
and	O
(	O
a	O
four-fingered	O
forked	O
hand	O
)	O
,	O
are	O
graphic	O
variants	O
of	O
a	O
single	O
glyph	B-Application
,	O
which	O
also	O
attaches	O
to	O
or	O
replaces	O
the	O
arms	O
of	O
various	O
other	O
glyphs	B-Application
:	O
</s>
<s>
The	O
fact	O
the	O
two	O
hands	O
appear	O
to	O
substitute	O
for	O
each	O
other	O
in	O
all	O
these	O
pairs	O
of	O
glyphs	B-Application
when	O
the	O
repeated	O
phrases	O
are	O
compared	O
lends	O
credence	O
to	O
their	O
identity	O
.	O
</s>
<s>
Similarly	O
,	O
Pozdniakov	O
proposed	O
that	O
the	O
heads	O
with	O
"	O
gaping	O
mouths	O
"	O
,	O
as	O
in	O
glyph	B-Application
are	O
variants	O
of	O
the	O
bird	O
heads	O
,	O
so	O
that	O
the	O
entirety	O
of	O
Barthel	O
's	O
300	O
and	O
400	O
series	O
of	O
glyphs	B-Application
are	O
seen	O
as	O
either	O
ligatures	O
or	O
variants	O
of	O
the	O
600	O
series	O
.	O
</s>
<s>
Despite	O
finding	O
that	O
some	O
of	O
the	O
forms	O
Barthel	O
had	O
assumed	O
were	O
allographs	O
appeared	O
instead	O
to	O
be	O
independent	O
glyphs	B-Application
,	O
such	O
as	O
the	O
two	O
orientations	O
of	O
his	O
glyph	B-Application
27	O
,	O
the	O
overall	O
conflation	O
of	O
allographs	O
and	O
ligatures	O
greatly	O
reduced	O
the	O
size	O
of	O
Barthel	O
's	O
published	O
600-glyph	O
inventory	O
.	O
</s>
<s>
By	O
recoding	O
the	O
texts	O
with	O
these	O
findings	O
and	O
then	O
recomparing	O
them	O
,	O
Pozdniakov	O
was	O
able	O
to	O
detect	O
twice	O
as	O
many	O
shared	O
phrases	O
,	O
which	O
enabled	O
him	O
to	O
further	O
consolidate	O
the	O
inventory	O
of	O
glyphs	B-Application
.	O
</s>
<s>
By	O
2007	O
,	O
he	O
and	O
his	O
father	O
,	O
a	O
pioneer	O
in	O
Russian	O
computer	O
science	O
,	O
had	O
concluded	O
that	O
52	O
glyphs	B-Application
accounted	O
for	O
99.7	O
%	O
of	O
the	O
corpus	O
.	O
</s>
<s>
|	O
colspan	O
=	O
13|	O
Glyph	B-Application
was	O
first	O
proposed	O
by	O
Pozdniakov	O
.	O
</s>
<s>
The	O
inverted	O
variant	O
27b	O
in	O
Barthel	O
's	O
glyph	B-Application
27	O
(	O
)	O
appears	O
to	O
be	O
a	O
distinct	O
glyph	B-Application
.	O
</s>
<s>
Although	O
looks	O
like	O
a	O
ligature	O
of	O
and	O
statistically	O
it	O
behaves	O
like	O
a	O
separate	O
glyph	B-Application
,	O
similar	O
to	O
how	O
Latin	O
Q	B-Algorithm
and	O
R	B-Algorithm
do	O
not	O
behave	O
as	O
ligatures	O
of	O
O	O
and	O
P	B-Algorithm
with	O
an	O
extra	O
stroke	O
,	O
but	O
as	O
separate	O
letters	O
.	O
</s>
