Gobioid Fishes — Gosline 
167 
lines of specialization as well as a major 
grouping around the typical gobies, With 
the meager amount of material at hand it 
seems impossible to delimit these adequately 
or properly. About all that can be said is that 
a division between the gobioid fishes with 
ventrals separate and those with ventrals 
united may be convenient, but that it prob- 
ably does more to obscure than to clarify the 
true phylogenetic sequences within the Go- 
bioidei. 
KRAEMERIIDAE 
As noted by numerous authors, e.g., Schultz 
(1943: 262), the genus Kraemeria has been 
described at least three times, i.e., as Kraemeria 
(Steindachner, 1906: 41), and Vitreola (Jordan 
and Seale, 1906: 393) from Samoa, and as 
Psammichthys (Regan, 1908: 246) from the 
Seychelles. The genus has been placed among 
the gobioids by Jordan and Seale (1906: 393), 
Regan (1911: 733), etc., and with the tricho- 
notids by Regan (1908: 246), Fowler (1938: 
300), Schultz (1943: 262), etc. A family 
(Psammichthyidae) was first erected for it by 
Regan (1911: 733). 
Aside from Kraemeria only two other gen- 
era seem to have been attributed to the 
Kraemeriidae. One of these is Gobitrichonotus 
Fowler (1943: 85). This genus would seem 
to differ from Kraemeria chiefly in that the 
pelvics are fused and the two dorsal fins are 
completely separate. In these features Gobi- 
trichonotus is intermediate between the typical 
gobies and Kraemeria , thus making the Krae- 
meriidae more difficult to define. 
More recently Whitley (1951: 402) has de- 
scribed the rather nondescript genus Par krae- 
meria as a kraemeriid. Whether or not this 
fish is actually related to Kraemeria I am 
unable to judge. 
The question of the rank to be attributed 
to Kraemeria and Gobitrichonatus within the 
Gobioidei cannot be finally determined until 
more is known about the other forms of the 
suborder. Tentatively, and in part because it 
causes no nomenclatorial innovation (cf. 
Berg, 1940: 487), the two genera Kraemeria 
and Gobitrichonotus may be recognized as a 
family Kraemeriidae. The family may, per- 
haps unsatisfactorily, be distinguished from 
other gobioid families as follows: 
Scaleless, small-eyed gobioid fishes with a 
projecting chin. Gill openings extending 
rather far forward, the gill covers narrowly 
attached to the isthmus. Pectoral rays 8 or 9- 
Dorsal and anal free from the caudal. Pelvics 
separate or united. Dorsals separate or united. 
If Par kraemeria actually belongs to the 
Kraemeriidae several of the diagnostic fea- 
tures given above will have to be omitted, 
leaving a rather weak residue, for Parkraemeria 
lacks a projecting chin and has large eyes 
and 15 pectoral rays. 
MICRODESMIDAE 
The microdesmids, to my knowledge, have 
been universally treated as blennioid fishes, 
and have been accorded family status, c.f. 
Regan (1912: 274). The Microdesmidae con- 
sists, according to Reid (1936), of a single 
genus Microdesmus described by Gunther 
(1864: 26) containing 10 species from both 
coasts of tropical America and from the 
Cameroons. However, certain species from 
the tropical western Pacific not included by 
Reid apparently also belong with the micro- 
desmids. In 1858 Bleeker described the genus 
Gunnellichthys from the East Indies. Jordan 
(1923: 233) placed this genus in his provi- 
sional family Chaenopsidae, and de Beaufort 
(in Weber and de Beaufort, 1951: 447) allo- 
cated it to the Pholidichthyidae. It appears, 
from evidence presented below, that Gunnel- 
lichthys must not only be extracted from tin* 
above families but removed from the blen- 
nioid fishes altogether and placed with 
Microdesmus among the gobioids. 
Kendall and Goldsborough (1911: 324) 
described the genus Paragobioides , based on 
P. grandoculis from the Marshall Islands. This 
species has been allocated to the percoid fam- 
