166 
PACIFIC SCIENCE, Vol. IX, April, 1955 
body and in the apparently correlated in- 
crease in number of vertebrae and change in 
dorsal fin structure. In the typical gobioids 
the vertebral number is apparently around 26, 
but in some of the more elongate species the 
number reaches 34. Microdesmus , with 42 to 
62 vertebrae seems to be simply the terminal 
point in a line of gobioid specialization for 
which intermediate steps are known. The 
absence of a separate spinous dorsal fin is 
also known in other gobioid genera, most of 
them again elongate forms; in many of these 
the dorsal spines are poorly differentiated 
from the soft rays. 
It may be well, at this point, to summarize 
the salient characters which would seem to 
distinguish the Gobioidei, including Krae- 
meria and Microdesmus , from other suborders 
of the Perciformes. 
Parietals lacking. Branchiostegals (4) 5 or 
6, the first one or two well separated from 
the others. Mesopterygoid narrow or absent. 
Preopercle and symplectic widely divergent 
above, with an interspace between them. Hy- 
purals with a splint-like bone above and 
below. 
A host of less diagnostic characters might 
be added. Nevertheless, it is necessary to weed 
out of the existing diagnoses of the gobioid 
fishes, for example those of Regan (1911: 729) 
and Koumans (in Weber and de Beaufort, 
1953: 1), a number of the features usually 
listed. 
The pelvic fins may be reduced ( Micro- 
desmus) beyond, "4 or 5 soft rays.” The epio- 
tics are not always ( Kelloggella , Microdesmus , 
and Kraemeria), "separated by the supraoc- 
cipitals.” The opisthotic is not always large 
{Eviota, Kelloggella , Kraemeria ). The lower 
pharyngeals are not always separate ( Awaous , 
Kraemeria). The posttemporal is not always 
forked ( Kelloggella , Kraemeria). There may be 
only three actinosts ( Kraemeria ). Finally, the 
number of vertebrae may be very much greater 
than 34 ( Microdesmus ). 
Before attempting to establish the position 
of Microdesmus and Kraemeria within the Go- 
bioidei it may be well to list in phylogenetic 
key form the principal osteological differ- 
ences between the various forms here ex- 
amined. 
la. Branchiostegal rays 6. . Eleotris 
lb. Branchiostegal rays 5. 
2a. Two or three soft dorsal interneurals to 
each neural spine ...... .Ptereleotris 
2b. Usually one soft dorsal interneural to 
each neural spine. 
3a. Fewer than 30 vertebrae; a separate 
spinous dorsal fin of 5 or 6 spines. 
4a. Four actinosts and 12 or more 
pectoral rays; gill membranes 
broadly attached to the isthmus. 
5a. No crest running along the 
whole top and back of head; 
branchial arches without or with 
weakly-developed gill rakers. 
6a. Lower pharyngeals separate; 
sphenotics meeting the su- 
praoccipital. 
7a. Epiotics separated by the 
supraoccipital; pelvic s sep- 
parate. Eviota 
7b. Epiotics meeting behind 
the supraoccipital; pelvics 
united Kelloggella 
6b. Lower pharyngeals fused; 
sphenotics not meeting the 
supraoccipital ..... Awaous 
5b. A crest running along the top 
and back of skull; branchial 
arches with two rows of spinous 
gill rakers Gobiodon 
4b. Three actinosts and 8 pectoral 
rays; gill membrane narrowly 
united to the isthmus .Kraemeria 
3b. Sixty- two vertebrae; no separate 
spinous dorsal fin Microdesmus 
This synopsis indicates the general opinion 
that in the Gobioidei Eleotris is undoubtedly 
nearest to the basic percoid stock from which 
the gobioids appear to have arisen. Beyond 
Eleotris there would seem to be a number of 
