2 10
B. J. MARPLES, F.L.S.
refers to the three pairs as ‘inferior’ ( =al), ‘median’ ( = p m ) and ‘superior’ ( =pl).
Savory (1928) refers to the ‘anterior or superior’ (=al), the ‘middle’ ( = p m ) and the
‘posterior or inferior’ (=pl). Some authorities homologize the spinnerets with biramous
appendages, and so the median spinnerets are referred to as ‘endopodites’ and the lateral
ones as ‘exopodites’. The question of the homology will be discussed later, and in the
meantime these terms will not be used. Berland (1932) stated that the four pairs of
spinnerets belong to four successive segments, but, as pointed out by Petrunkevitch
(1942), this was ‘due to an oversight or a misunderstanding’ and is nowhere supported.
A convenient separation of the order Aranea is into three suborders, the Liphistiomorpha, the Mygalomorpha and the Araneomorpha. These will be considered separately.
LIPHISTIOMORPHA
This suborder contains the single Recent family Liphistiidae, with the two genera
Liphistius and Heptathela. These are regarded as the most primitive living spiders, and
the condition of the spinnerets is part of the evidence for this. The spinnerets are situated,
not a t the posterior end as they are in most spiders, but about the middle of the ventral
surface of the abdomen, close behind the second pair of lungs and far removed from the
anus. Both pairs of lateral spinnerets differ from those of other spiders in being manyjointed. They are conical in shape and divided by grooves into up to about 12 joints, of
which the basal one is much the largest. They are covered with setae amongst which are
the fusules by which the silk glands open. The median spinnerets are much smaller and
are single-jointed. I n the genus Liphistius there are eight spinnerets, the size of the small
pairs of am and pm varying slightly in different species (Text-fig. 1 ) .
Millot (1932) tabulates the number of glands associated with the different spinnerets
among the liphistiomorphs. The a1 always have the greatest number, and the pl the
next largest, there being 200 and 140 respectively on these spinnerets in L.desultor. Of
the median spinnerets the um are functionless in all seven species of liphistiomorph
examined. The p m are functionless in four species and have only one to four glands
associated with them in the other species.
I n Heptathela the p m are united into a small median structure which has no silk glands
associated with it (Text-fig. 2). It shows slight traces of its double origin in the arrangement of setae and in having an internal septum. Yoshikura (1955) has shown that
Heptuthela when first hatched has two pairs of spinnerets, one situated on the fourth
and one on the fifth abdominal segments. After the first moult there are eight spinnerets,
but the posterior median ones subsequently fuse together. Millot (1949) gives a table t o
show the arrangement of spinnerets in different groups of spiders. I n this i t is stated
that it is the a m which are fused in Heptathela. This must be a lapsus calami as there seems
no doubt that it is the p m which are fused, as indeed was pointed out by Millot himself
(1932). As will be discussed below, in araneomorphs the am may be united into a median
vestige called the colulus, and what we see in Heptathela is a posterior colulus, not a
median spinneret as was once supposed. Millot (1932) points out that among the living
liphistiomorphs we see the median spinnerets on the verge of disappearance. The posterior
colulus of Heptathela varies in size between individuals, being sometimes scarcely visible,
and he says, ‘I1 ne serait nullement suprenant que l’on v h t un jour A dbcouvrier des
Liphistiides hexathAle, par disparition complete des filihres post6ro-internes, ou m6me
pentatheles, par fusion concoinitante des filieres ant6ro-internes, en un rudiment unique ’ .
Simon (1897) established a genus Anadiasthothele for a species of liphistiomorph from
Sumatra, and Kishida (1923) placed i t in a separate subfamily. The main characteristic
was that the spinnerets were single-jointed. Re-examination of the type by Bristowe
(Bristowe & Millot, 1932) showed that the spinnerets were actually many-jointed, and the
species is referable to the genus Liphistius.
211
Xpinnerets and epiandrous glands of spiders
om
-am
-aI
- Pm
-P'
4
1
7
om
co1.
a1
5
6
2
B
U
3
9
Text-figs 1 to 9. Spinnerets of certain spiders in ventral view, with all setae and fusules
omitted. Xot t o scale.
i . Liphistius bafueiasis, Liphistiiclae (Maiaysia). am and p n present but non-functional
(after Bristowe). 2. Heptathela ki)).l.urni, Liphistiidae (Japan). p m fused into a posterior
colulus. 3. Hezuthele petriei, Dipluridae (New Zealand). am absent altogether. 4. Filistato
bakeri, Filistatidae (Samoa).a m form a divided crihellum. 5. l a e t ~ t i c u martius,
s
Amaurobiidae
(New Zealand). am form a divided crihellu~ii.6. Hypochilus thorelli, Hypochilidae (United
States). am form an undivided crihellum. 7. Agelenid (Uganda). am vestigial and non-functional. 8. Cambridgea sp. Agelenidae (New Zealand). am fused t o form a broad anterior
colulus. 9. Loxosceles sp., Loxoscelidae (Uganda). am fused to form an elongated anterior
eolulus.
MYGALOMORPHA
This suborder contains a large number of species divided among some eight families.
They are often referred to as the trapdoor spiders, though not all make burrows with
this kind of a n opening, or as the tarantulas. The majority of them have only two pairs
of spinnerets, but a small number have three pairs and a few a single pair only. More
differences of opinion have arisen over which spinnerets are retained by mygalomorphs
than over the situation in the other groups. Savory (1928) states that in mygalomorphs
with two pairs of spinnerets these are a1 and pm, while pl are occasionally present. This
is not in accord with the opinion of the great majority of authors, and the confusion seems
to have arisen over the use of the terms 'superior' and 'inferior'. I n some instances, as
in human anatomy, these are taken to mean cephalad and caudad, in others the meaning
is topographical, upper and lower with reference to the ground. I n spiders the lateral
212
B. J. MARPLES,F.L.S.
spinnerets of the fourth abdominal segment are the nearest to the head and so might be
termed ‘superior’, but, as they are the nearest to the ground they also might be termed
‘inferior’. Simon (1892) refers to the small median spinnerets of mygalomorphs ( p m ) as
‘inferior’ because of their position, though he does not regard them as corresponding to
the ‘inferior’ ones of araneoniorphs, which are al.
The opinion of the majority of authorities, and there seems no doubt that it is the
correct one, is that in mygalomorphs with two pairs of spinnerets these are pl and pm,
and that in the few mygalomorphs which have three pairs these are al, pl and prn. am
are never present in mygalomorphs, even as a vestige, (Simon, 1892; Petrunkevitch,
1933; Gerhardt & Klistner, 1938; Millot, 1949; Gertsch, 1949). The study made by
Montgomery (1909) on the development of a mygalomorph (Evagrus,family Dipluridae)
confirms this conclusion based on adult structure. The fourth abdominal segment of
Evagrus has transient rudiments of appendages, while those of the four spinnerets appear
on segment 5. Yoshikura (1958) showed that in Atypus the anterior spinnerets arise
on segment 4,while the other two pairs arise on segment 5.
A very differcnt opinion with regard to the mygalomorphs was offered by Petrunkevitch
in 1942. He was considering evolutionary trends in spiders and gave very useful tabular
diagrams to illustrate 36 arrangements of respiratory organs, heart and spinnerets. He
said, ‘In Mygalomorphae the posterior median spinnerets are never found. They must
have disappeared even before the suborder came into existence. Of the remaining six
spinnerets the anterior median ones are the first to disappear to be followed by the
anterior lateral ones, i.e. by the exopodites of the pleopods of the fourth somite. The
posterior lateral spinnerets persist in all mygalomorph spiders. The rule, then, for spiders
with paraxial chelicerae is that the posterior median spinnerets dixappear first, then the
anterior median and finally the anterior laterul ones’. This seems to be a deduction from
the fact that in Heptathela the pm are vestigial, being combined into the posterior colulus.
As the quotation from Millot, given above, emphasizes, it is easy to imagine this process
being continued and the pm and even the am being lost altogether in the liphistiomorph
stock. I n the mygalomorphs however, all the evidence points in a different direction, and
in this group it is the am which are always absent, and, when only four spinnerets are
present these are pl and pm. This also was Petrunkevitch’s opinion in 1933.
There are differences of opinion as to the number of joints present in the spinnerets
of mygalomorphs, though undoubtedly p l usually have three joints, while pm are singlejointed. According to Simon (1892) a n exception to this is Missulena (Ctenezidae) in the
female of which the inferior spinnerets (pm) are distinctly two-jointed, while they are
single-jointed in the male. I have examined a specimen of Missulena from West Australia
in which pm are single-jointed. Masteria hirsuta (Dipluridae) also was described and
figured, by Koch (1871), as having the p m two-jointed, and I can confirm this from a
specimen in my own collection. Occasionally the p1 are four-jointed, a condition known
in a few genera of Dipluridae, Ctenezidae and Theraphosidae.
The majority of mygalomorphs have four spinnerets only, which are pol ancl pm, but
a few have six, in which case a1 are also present. Several families, together with the
Liphistiidae, were grouped by Gertsch (1949) as the ‘atypical tarantulas’. I n AZiatypus
a1 are nearly equal in size to pm and are two-jointed, and they are functional, as shown
by the presence of well-developed spigots. He says that this ‘marks them as the most
generalized of all mygalomorph spinnerets except those of the Liphistiidae ’. In Atypoides
and some other genera a1 are unsegmented and small, while in Antrodiaetus they are absent.
These genera belong to the family Antrodiaetidae. Hexura, in the Mecicobothriidae, has
six spinnerets. the pl being extremely long. They spin sheet-webs, similar to those of the
Dipluridae in which family they are sometimes placed. The Atypidae also have six
spinnerets, a1 and p m being small and single-jointed, p l being larger. In these ‘atypical
tarantula’ families the spinnerets are more or less remote from the anal tubercle, a
condition somewhat resembling that found in Liphistius.
Spinnerets and Ppiandrous glands of spiders
213
Among the .typical tarantulas', some members of the family Dipluridae, the genera
Ilexdhele and Scotinoecus, have six spinnerets. It is characteristic of this family that the
pl are very long and three-jointed, usually being used for spinning a dense sheet-web.
This behaviour is regarded as a character of the family, but some species of Hexathele
in New Zealand dig deep burrows in the ground, lined with silk but not closed by a trapdoor. These burrows are definitely excavated by the spider, not merely crevices inhabited
by it. The silk lining usually stops at the entrance and there is no external sheet, though
a few threads may extend out t o the surrounding vegetation. There has been controversy
over the number of joints in a1 of Hexathelr, and Machado (1944) in discussing this supported the wrong conclusion. I-lexathele has ril which have two joints, the terminal ones
being small. They are functional and slightly larger than the pm, which are single-jointed.
p l are very long and are three-jointed. Hexathd(>therefore has spinnerets as generalized
as those of Aliatypus mentioned above (Text-fig. 3 ) . I n the related genus Yorrothele
the u hole appearance is very similar to that of Hexathele but the a1 are absent altogether.
I n a few mygalomorphs only two spinnerets are present, and, though their structure
and development do not seem to have been studied, it seems certain that they are p l .
Some members of the families Paratropiclitla and Barychelidae have this arrangement.
hlachado (1944) shoved that Nemesia hzspa ca has only two spinnerets (pZ), this being
the only knon n example of variation in the number of spinnerets with a genus. It is also
the only member of the Ctenezidae with only two spinnerets.
ARANEOMORPHA
The great majorit>yof araneoniorph spiders have three pairs of functional spinnerets,
which are undoubtedly al, p l and pm. Some of the Zodariidae, belonging to the suborder
CyPrelinae, have four spinnerets which, as they each have two joints, are probably al,
and pZ. The same seems to be true of the Hadrotarsidae, though there is some doubt
about, the number of spinnerets which they have (&lachado,1944).Some other zodariids,
some of the Archaeidae and the Palpirnanidae have only a single pair of spinnerets,
which appear to be bhe al. This is shown by the fact that there are related genera in
which t.he a1 are large, and the p1 and p m show various degrees of reduction. Machado
(1941) makes the very interesting observation on Palpimanus gibbulus, that between
t,he single pair of spinnerets and the anal t,ubercle is a clearly marked membraneous area
carrying large fusules. He figures 3G fusu1c.s in this area in an adult female, and only six
on each spinneret. This presumably represents the regression and modification of one or
both pairs of posterior spinnerets. I n yowig specimens there are few fusules, grouped
a t each end of the area and indicating it,s paired origin. This transformation of paired
projecting spinnerets into a flat spinning area is especially interesting in view of the
formation of the cribellum from the um, which has taken place in several families of
araneomorplis. Another unusual example of spinnerets being shortened and the spinning
field broadened out into an area carrying numerous fusules is afforded by Rebilus
(Drassidae) described by Simon (1892). Here hlie pm are drawn out into two elongated
areas lying longitudinally between the bases of t'he other four spinnerets and each provided
with two regular lines of large short fusules.
Almost invariably the a1 and pl have tJwojoints and the pm are single-jointed. I n the
Pholcidae the p l are single-jointed. A peculiar situation exists in Ammoxenus. Here there
is a pair of posterior spinnerets in front. of which is a row of four of which those nearest
to the middle line have two joints, while the outer ones have only one. Petrunkevitch
(1933) interprets this as an abnormal location of spinnerets, aZ being median to pm.
Machado (1944) and Simon (1892) on the other hand interpret it is abnormal jointing
of the spinnerets, pm being median but having two joints instead of the usual one while
a1 have one joint instead of two. The most familiar example of unusual arrangement of
spinneret's is that found in Hahnia, where all six spinnerets are arranged in a transverse
B. J . MARPLES,
F.L.S.
214
line. Here it is clear that the sequence, from outside inwards, is pl. al, pm. Occasionally,
as in some Drassidae and Prodidomidae, the a1 are separated from the other spinnerets
and arise further forward along the ventral surface of the abdomen. I n Gryptothele
(Zodariidae) the group of spinnerets is capable of being retracted below the surface of the
abdomen, so that all that is then visible is a slit surrounded by setae.
The spinnerets may be more or less equal in length, though the p are the smallest,
but sometimes some are conspicuously elongated. This elongation reaches its maximum
"
U U 1
(d)
Liphistiornorpha
* '
-b'
'I
I ''I
(g 1
Mygalomorpho
Araneomorpho
I''I
..
* .
(nl
Ecribellata
Text-fig. 10. Diagram showing the arrangement of the spinnerets in different spiders. Those
shown above the horizontal line are on the fourth abdominal segment, those below the l i e
on the fifth. No attempt is made to show the size or structure, but when spinning glands are
present the spinneret is shown in solid black. The dots in the mygalomorph and ecribellate
columns indicate the complete absence of certain spinnerets.
( a )Hypothetical ancestral stage with eight functional spinnerets. ( b ) Liphistius desultor. a m
not functional. ( c ) Liphistius batuensis. a m and p m not functional. ( d ) Heptalhela. a m not
functional, pm fused t o form a posterior colulus. ( e ) Hexathele. a m absent. ( f ) Commonest
mygalomorph condition. al and am absent. ( 9 ) Diplothele. Only pl present. ( h ) Filwtata. a m
partly fused to form a divided cribellum. ( i )Dictyna. am fused to form a n undivided cribellum.
(j)Agelenid. a m vestigial and not functional. (k)Common araneomorph condition. a m fused
t o form anterior colulus. ( 2 ) Common araneomorph condition. am absent. ( m ) Gmogala. am
and p m absent. (n)Mecysmauchenius. Only al present.
Spinnerets and epiandrous glands of spiders
215
in the Hersiliidae, where the p l , especially the end joint, are as long as the abdomen.
They have been cited as an example of large spinnerets associated with small spinning
ability, as these spiders do not make a web. It has been shown however (Bristowe, 1930)
that this is a specialization for catching prey, by running round it with extreme rapidity
while emitting sheets of fine silk from the long spinnerets. The Agelenidae characteristically have long p l , a condition which is associated with spinning a sheet-web, as it is
also in the Dipluridae.
The truncated end of the spinneret carries the fusules in the spinning field. The number
of silk glands and the size of the fusules by which they open on the spinnerets is very
varied. According to Millot (1930) the p l of Pholcus and the pm of Loxosceles are functionless, while the pl of Scytodes have only one fusule. Numbers of fusules per spinneret for
one of the Argiopidae have been given as al, 70: p l , 120; pm, 150.
Among the araneomorph spiders the six spinnerets, al, p1 and pm, show only these
minor differences in different families. They are also similar to those of the mygalomorphs,
though it is noteworthy that when they are reduced to a single pair this is the pl in the
mygalomorphs while i t is the a1 in the araneomorphs. am are altogether absent in the
mygalomorphs, but they are represented in the araneomorphs and their condition is a
matter of considerable interest.
Broadly speaking there are three possibilities. The am may be present as functional
spinning organs, they may be present as functionless vestiges or they may be absent
altogether. When they are functional as spinning organs they have the form of a transverse plate, more or less divided into two halves, lying anterior to the group of spinnerets.
This is the cribellum. It is covered with a very large number of minute openings for
special silk glands which occupy the posterior ventral part of the abdomen. Associated
with the cribellum is a pair of combs of setae, called the calanzistra, carried on the metatarsi
of the posterior legs. To produce cribellated silk a calamistrum is moved rhythmically
backwards and forwards across the cribellum, combing out its secretion and depositing
it on two or more stout threads spun simultaneously by the spinnerets. Among the
araneomorph spiders 12 families possessing a cribellum are commonly recognized.
Of the remaining araneomorph families probably between half and two-thirds of them
have a vestigial structure called the colulus, lying in front of and between the ul. No
extensive study of the internal structure of the colulus seems to have been made. Millot
(1930) examined those of several of the Scytodidae, a family in which the colulus is
unusually large, and found within simply a blood-space and no structures suggesting
that there was any function. The structure of the colulus was, in fact, identical with
that of a non-functional spinneret. There seems to be no doubt that the colulus is a
vestige of the um which have become united in the middle line, in exactly the same way
that the prn of Heptatheln have given rise to a posterior colulus. Traces of the double
origin may still be discernible in the colulus, as an internal septum or in the external
form or setation.
The most common form of colulus is a small triangular or conical lobe, which, a t its
largest as in the Scytndidae, may be about half the length of the a1 (Text-fig. 9). Many
stages in the reduction of the conical form may be seen. Sometimes there may be a low
mound, or a plate of thickened cuticle in the corresponding place, sometimes merely a
pair of stout bristles, as in some Oonopidae, or even a single one as in some Pholcidae
(hfachado, 1944). Sometimes, as notably in some Agelenidae, the colulus is broader
than long and has the form of a transverse ridge or lobe (Text-fig. 8). It may be represented by two distinct seta-bearing plates, strongly suggestive of a vestigial pair of
am (Text-fig. 7). The first example of this was described by Montgomery (1909) in
Agelena naevia.
In some families the colulus is always present, while in others some members are without
it. No doubt such a vestigial structure could easily be lost, and its presence or absence
will be of little help in determining relationships
216
B. J.MARPLES,F.L.S.
In having a colulus, which is a vestige of the am, these araneomorph spiders are more
primitive than the mygalomorphs, none of which have any trace of this pair of spinnerets.
The cribellate araneomorphs have a functional representative of the anterior median
pair, and so, in this respect, are the most primitive of all living spiders. Even the liphistiomorphs which have a pair of am, never, a t the present day, have spinning glands opening
on them.
The cribellum consists of a transverse, well-marked plate lying in front of the group
of spinnerets. It may apparently be tilted by muscles attached, like those of the spinnerets, to one of the abdominal endosternites, but it usually lies flush with the surface
of the body. Occasionally, however, as in Hypochilus, it is distinctly elevated (Text-fig. 6).
The cribellum is described as being either divided or entire, but this is a matter of degree.
Sometimes the spinning area is clearly double, (Text-figs 4 and 5) sometimes it is
divided by a fine longitudinal line and sometimes there is merely a notch in the middle
of one margin. I n the Eresidae the spinning area of the cribellum may be divided into
four more or less equal areas, or it may have two main areas each with a row of five small
ones behind it.
Montgomery (1909) studied the development of the cribellum and found that it arose
from paired thickenings lying median to the appendages of the fourth abdominal segment.
The appendages become the al. The median thickenings become either the cribellum or
the colulus, and there seems no doubt that these are the same as the am of the liphistiomorph. Montgomery examined the cribellum of the adult Filistatn, which belongs to the
family in which the division of the cribellum is perhaps the most complete. His conclusion
was that ‘the spinning plates are merely the free, uncovered apices of a pair of spinnerets,
the greater region of which lies covered by the cuticular frame. Or the relation might be
represented in another way : the cribellum of Filistata is a pair of spinnerets medially
approximated, their glands opening distally on the spinning plates, and across and
around these spinnerets has developed a cuticular frame ’.
Assuming that the ancestral araneomorph had eight functional spinnerets it is possible
that in one line of descent these lost their function and became a colulus, while in another
they became the cribellum (Text-fig. 10).Following this. some systematists have divided
the suborder Araneomorpha into t M o divisions, the Cribellata and the Ecribellata.
Resemblances between representatives of these, which are a t times striking. are regarded
as due to convergence. An analogous case would be the resemblances between certain
of the marsupial and placental mammals.
Another phylogenetic possibility is that the a m of the ancestral araneomorph became
transformed into a cribellum, and that this has subsequently been lost in the majority
of families. As evidence for this mas cited the condition of the Hypochilidae, which are
more primitive than other araneomorphs in having four lungs and are also cribellate.
The weight of this evidence has been reduced by the recent discovery of Gradunguln,
an araneomorph with four lungs, but with no cribellum and a scarcely distinguishable
colulus. Even if the group of Cribellatae is a natural one home of its members may have
lost the cribellum, and so the ecribellate condition may have arisen by more than one
route. The situation is complex and controversial, and this is not the place to explore it.
H01\20LOGIES
The spinnerets are carried on two of the abdominal segments of the spider : the a1 arid
the am or cribellum or colulus on the fourth segment, the p1 and pnz on the fifth. By
many authors the lateral spinnerets are referred to as exopodites and the median ones
as endopodites. Petrunkevitch (1942) saps, “like all abdominal appendages of arthropods
they are true pleopods, biramous in structure’. And Kaston (1948), ‘the spinnerets
probably represent the homologues of the biramous pleopods of Crustacea ’. Machado
(1944) demonstrated that in a number of families of araneomorphs the adjacent lateral
Xpinnerets and epinndyous glands of spiders
217
and median spinnerets tend to arise on a common membraneous base, which he regards
as the protopodite of a biramous limb.
To call the spinnerets of spiders biramous limbs, and to compare them with the pleopods
of Crustacea, seems to be taking too general a view of the arthropod classes. The matter
of the relationship between these classes, and the comparison between their structure
is highly speculative and controversial. Recent tendency is to exclude the Onychophora
and related groups from the phylum Arthropoda, and to divide this into the Proarthropoda
including the trilobites, and the Euarthropoda (Vandel, 1949). The Euarthropoda are
subdivided into the Chelicerata, which includes the Merostomata, Arachnida and
Pycnogonida, and the Mandibulata, which includes the Crustacea, Myriapoda and
Insecta. The distinction between the Chelicerata and Mandibulata is so clear that some
authorities have maintained that they had separate evolutionary origins.
The biramous limb is characteristic of the Crustacea and does not occur in the other
classes. It consists of a basal, two-jointed protopodite from which arise the two branches.
The ventral endopodite is the main branch of the limb, the lateral branch is the exopodite.
A4notherview of the arthropod limb is that it is an elongated jointed appendage whose
basal part may carry projections, endites, 011 the median side, and exites or epipodites
on the lateral side. The exopodite of the crustacean biramous limb would thus be a large
epipodite on the second joint, while the endopodite would be the main axis of the limb,
the telopodite. The trilobites, a very primitive arthropod group, have limbs which have
two branches but whose structure is not regarded as being the same as the biramous
limb of the crustacean (Stormer, 1949). The outer branch of the trilobite limb seems to
be a pre-epipodite, probably with respiratory functions. The separation between the
two divisions of the Euarthropoda is so clear that it is impossible to regard the spinnerets
of spiders as corresponding to the pleopods of Crustacea.
The origin of the main divisions of the Arthropoda, the Chelicerata and Mandibulata
is still very obscure, and the enigmatic assemblage of forms from the Middle Cambrian
Burgess Shales does little to elucidate it. It seems that a possible ancestor might have
resembled a trilobite, with antennae and a long series of similar limbs carrying respiratory
pre-epipodites. To transform such a form into a merostome chclicerate would involve
the loss of the antennae, and the absence of the deutocerebrum in the chelicerate brain
supports this. The first pair of appendages would have to be converted into chelate
chelicerae and the next five pairs into walking legs with loss of the pre-epipodites. The
flabellum carried on the fifth leg of Xiphosura seems to be a reduced epipodite, and
vestiges have been described in the embryos of scorpions. Starting with a reduced segment,
the opisthosoma carries appendages wliicli have a jointed telepodite and a many-jointed
respiratory pre-epipodite. This is still seen in the living Xiphosura, where the first pair
of opisthosomatic appendages is fused t o form the genital operculum, and the remaining
five pairs carry gills on many-jointed epipodites.
The Arachnida differ from the Merostoniata in being without gills. and usually without
opisthosomatic appendages. The exceptions are the scorpions, whose pectines are modified
appendages of the third opisthosomatic segment, and the spiders, whose spinnerets are
on the fourth and fifth segments. Studies on the development of spiders by Montgomery
(1909), Holm (1940) and others seem to show conclusively that thc lateral spinnerets
represent the appendages of these segments. It is not surprising that in the early stages
of development of an arthropod the rudiments of appendages appear on many of the
segments, whatever their fate in the later stages of ontogeny. I n the opisthosoma of
arachnids this is true, and, according to Danydoff (1949) the largest number is nine to
ten pairs in Solpugida, while in the Pedipalpida and Acarina they seem to be absent.
Usually they disappear during development. The median spinnerets of spiders have
been shown by Montgomery to develop from thickenings of the ventral surface median
to the rudiments of the appendages which n 111 become the lateral spinnerets. He states
this specifically as previous workers had described the appendages as bifurcating,
218
B. J. ~MBRPLEs, F.L.S.
Montgomery did not agree with this. Yoshikura however states that in Heptuthela (1955)
and Atypus (1958)the median spinnerets arise by the bifurcation of the limb rudiments.
The details of early development are not necessarily conclusive in deciding the phylogenetic history of an organ. The lateral spinnerets are almost always better developed and
jointed, as compared with the small single-jointed median ones. We have seen that we
cannot equate the spinnerets of spiders with the biramous pleopods of Crustacea, and
they seem to have very little in common with the opisthosomatic appendages of the
Merostomata.
That two opisthosomatic segments of the spider retain appendages modified as spinnerets is not a t first surprising, since all arthropod segments potentially carry a pair of
appendages. When one looks a t the arachnid orders however, they are seen to be unusual
Petrunkevitch (1949)recognizes 16 orders, of which five became extinct in the Palaeozoic.
The earliest fossils known are scorpions from the Silurian, and the geological record as
far back as the Carboniferous is a good one. None of these arachnids, except the scorpions
and the spiders, show any trace externally of opisthosomatic appendages, and the Aranea
is one of the most highly evolved of the orders.
Specimens of fossil spiders showing spinnerets have been described. Fritsch (1904)
mentioned their occurrence on specimens of Palaeozoic spiders, and in particular figures
Eopholcus pedatus with two pairs of long-jointed spinnerets. Petrunkevitch (1953),
however, re-examined the material and stated that no spinnerets a t all are distinguishable.
Many examples of spinnerets, similar to those of Recent spiders, are known from specimens
preserved in Oligocene amber (Petrunkevitch, 1942).
Let us for a moment consider the phylogenetic relationships of the insect as a comparison
with the spider. I n insects also, there has been a development of walking legs on anterior
segments and suppression of appendages on the abdomen, though the mandibulate
insect of course has an elaborate head with antennae and mouthparts, absent in the
Chelicerata. The origin of insects from a Symphyla-like ancestor has been demonstrated
by Tiegs (1947)and others. Starting with a uniform series of segments mith walking legs.
the three thoracic segments with their large legs became distinct from the abdomen
where the appendages became reduced. In the Thysanura there are vestigial anterior
abdominal appendages, one or two pairs associated with reproductive activities and a
posterior pair of cerci. The more primitive pterygote insects still have the cerci, which
are eventually lost in the higher orders. All stages leading to the specialized limbless
abdomen of the advanced insect exist, and the evolutionary line is easily followed.
Though the arachnids also have a limbless opisthosorna, none of the stages leading to
this condition are known, even in the Palaeozoic. There is no doubt that the arachnid
orders arose long before the insects, but it is remarkable that no vestigial opisthosomatic
appendages have persisted.
Glands opening on appendages are of widespread occurrence, and so their presence on
the abdominal appendages forming the spinnerets is not surprising. Silk glands open on
the chelicerae of pseudoscorpions and poison glands on their pedipalps. Poison glands
open on the chelicerae of spiders, and in Scytodes they produce a silk-like material as
well as poison. Outside the Chelicerata mention may be made of the slime glands opening
on the oral papillae of Peripatus and the smaller crural glands on its legs, and the glands
which produce a silk-like material a t the tips of the cerci of Symphyla and of Anajapyx.
Since the median spinnerets do not appear to be appendicular in origin they offer a
more difficult problem. I n Symphyla there are vesicles adjacent to the bases of the legs.
stated by Tiegs (1947) to be independent of the legs and to be developed from the ventral
organ which is part of the ectodermal thickening which gives rise to the ganglia of the
central nervous system. Similar eversible sacs occur in the Thysanura, but here they are
on the bases of the appendages, as are the gonapophyses of insects on the appendages
of the genital segments. The Palpigrada is in some ways the most primitive of the arachnid
orders, but owing to their small size these animals may be simplified. They have three
Spinnerets and epiandrous glands of spiders
219
pairs of sacs formed of invaginated cuticle on the ventral side of the fourth, fifth and
sixth abdominal segments.
Ventral glands may also be present. In the Palpigrada there are glands opening on
bristle-bearing swellings on the vcntral side of the fourth and sixth abdominal segments
(Millot, 1949). Also of interest in this connection is the presence in male spiders of what
may be called epiandrous glands. These are rnulticellular glands whose ducts open on
liollow setae anterior to the genital opening, strongly recalling the structure of the
spinning glands. Their structure and occurrence are discussed below, their function is
unknown.
It seems therefore that appendages with the opening of glands may be expected to
occur. Also that glandular structures may be present on the ventral side of abdominal
segments and it is conceivable that the median spinnerets might have arisen from something of the sort. Two ways in which the structure of the spider might have evolved may
be imagined. I n the first, the ancestral line leading to the spiders must have diverged
from those of the other arachnid orders a t a stage when opisthosomatic appendages were
still present. I n the ancestral spider the two pairs of appendages and their glands, together
with the ventral glandular structures in the sarne segments became specialized into the
eight spinnerets. All the other arachnids lost these structures a t a very early period.
The other possible explanation is that early in arachnid history the division of the body
into prosoma and opisthosoma became established. and the opisthosornatic appendages
were suppressed in the adult though present in the early embryonic stages. I n the ancestral
spiders the ventral glands of the fourth and fifth opisthosomatic segments became
specialized as spinnerets. Associated with this emphasis on these segments, by a process
of paedomorphosis the limb-buds were retained into postembryonic life and in due
course became the lateral spinnerets. Their jointed leg-like form might be secondary,
and not the normal jointing of a limb. If this be true the median spinnerets would be the
original ones, joined in course of time by the lateral ones which have largely replaced
them.
Neither of these explanations is very satisfactory. The first involves a very early
separation of the spiders from all the other orders of arachnids, which seems unlikely.
The second involves complicated evolutionary manoeuvres of an improbable kind.
Neither has any palaeontological support a t present, and it is to be hoped that another
fortunate discovery such as the Burgess Shale or Rhynie Chert will be made. Both of
these provided an amazing amount of detailed information about early arthropods, and
something of the sort is badly needed to throw light on the early history of the
Chelicerata.
THE EPIANDROUS GLANDS
The name epiandrous glands seems appropriate for a group of well-developed glands
which open immediately anterior t o the genital furrow of male spiders. They seem to
have been first noticed by Machado (1951), who recorded that in the Ochyroceratidae
there is a row of three fusules, rarely two or four, anterior to the male genital opening,
similar in size and shape to the smaller fusules of the inferior spinnerets. He stated that
they were evidently homologous with those found in a similar position in Scytodes and
Pholcus, but that they were absent in males of Leptonetidae. I have examined specimens
of other species of Scytodes and Pholcus and found a situation similar to that figured by
Machado. Among the Ochyroceratidae I recorded (1955) the presence of six fusules in
Ceruleocera and two in Apiacera, species from western Samoa.
Machado compares the ‘genital fusules’ of spiders with structures seen in Koenenia,
and says that they might contribute to the semen or take part in the formation of the
sperm web. He supposed that they might represent the vestiges of appendages of the
second abdominal segment in the same way that the spinnerets represented those of the
B. J. MARPLES.F.L.S.
220
fourth and fifth segments. I n my view it is the lateral spinnerets which represent the
appendages of the fourth and fifth segments, the median spinnerets being ventral glandular
organs lying between their bases. I n the second segment the lungs to some extent represent
the appendages, while the epiandrous glands are the ventral glandular organs. They
therefore seem to be the serial homologues of the median spinnerets. I n the third segment
there are lungs in the tetrapneumone spiders only, and I have not been able to find any
glandular organs associated with them. Yoshikura (1954) records what he described as
a rudimentary tracheal organ, which has a median opening between the second pair of
lungs in Heptatheh. He states that it does not originate from the abdominal appendages.
During the studies on the internal anatomy of hypochilomorph spiders which led to
this consideration of spinnerets in general, attention was directed to the epiandrous
glands by the observation of their great size in Gradungula. Here the area between the
waist and the genital furrow, extending almost out to the lungs on each side, is occupied
by some 40 to 50 large glands. They are cylindrical or pear-shaped, with a large lumen
surrounded by a tall columnar epithelium. Their ducts run back and open to the exterior
by means of stout hollow setae or fusules. These arise on a transverse area on the anterior
border of the genital furrow. The fusules arise in sockets in the cuticle and are tapering.
but they do not have the cylindrical basal part and narrow distal part so often seen in
the fusules on the spinnerets. The ducts have a cuticular lining which extends right up
t o the gland and expands slightly where the lumen is reached. This is seen also in glands
opening on the spinnerets. Two individuals were examined and in neither did the glands
contain much secretion. It seems to turn hard on fixation in Alcoholic Bouin, and the mass
stains blue centrally and yellow peripherally in Mallory's stain. The silk glands, which
are very similar in size and appearance, seemed much more active, with the newly-formed
secretion in this case staining blue. Gradungula is a spider which makes very little use
of silk. It spins no web, though large stalked egg-cocoons are made a t least by caveinhabiting species.
Attention was given to specimens of males belonging to 26 families. Members of 11
were sectioned, of ten the cuticle was cleared by boiling in potash and mounting on a
slide, and six were only examined superficially. This last is not satisfactory for demonstrating the absence of glands, though a t times their presence is easily visible. The result
I
I
0.5m m
Text-fig. 11. Semidiagrammatic sagittal section of a male Hickmcctiicc, based on the s,-ries
of sections of which P1. 1 , fig. 4 is one. Anterior end to the left. The genital furrow is seen
leading up to the anterior abdominal endosternite, shown in black. On its anterior surface
is the genital opening (g.o.),the sperms in the genital duct being stippled. Below this, parts of'
seven epiandrous glands ( e . c c . q . ) are visible, one with a duct traversing t,he mass of unicellular
glands ( u . g . ) . Two fusules (f)of the epiandrous glands are shown. Above the endosternite is
the ventral diverticulum of the mid-gut, ( m g . ) and above that part of the silk glands (8.g.).
Spinnerets and epiandrous glands of spiders
22 1
of this survey is the conclusion that epiandrous glands are usually present in male spiders
though their extent may vary considerably.
Among the mygalomorphs sections sliowcd that the glands are present in Ctenezidae
and Dipluridae very much as in Gradungula described above, with the area occupied
by the fusules extending more anteriorly. I n a single specimen of the Theraphosidae the
presence of fusules was doubtful, but a raised pale area between the anterior lungs suggested the presence of glands. I n Antrodiaetus a small area anterior to the fold seemed
to have fusules.
In Hypochilus the glancls are pear-shaped and relatively smaller than in GradunguZa.
I n Iiickmania the glands are well developed (Text-fig. 1 1 ) . I n the area where the fusules
are located there are tall columnar glandular cells forming a mass, clearly enclosed by
a basement membrane and transversed by the ducts leading t o the fusules. The ducts
of the columnar glandular cells seem to open through the cuticle without any fusules.
Among the araneomorphs the glands secni t o vary from a condition where numerous
fusules occupy an area anterior to the furrou and the glands extend anteriorly some
distance, as in Pilistata and Dinopis,to one nliere there is a single row or merely a pair
of fusules, as in some Ochyroceratidae. Sometimes the fusules are grouped and their
bases lie in distinct pits in the cuticle. This is very noticeable in the Argiopidae, where
the pits are easily visible in the intact specimen. They were seen in members of three genera,
Aryiope, Araizeus and Cyrtophora. They vary between individuals, in A . pustulus one
individual had l’ipits, while two others had seven and nine larger ones. Sections discloscd
a condition similar to that in Hickmania, the ducts passing through a mass of elongated
unicellular glands. Members of the follov ing families, usually only one or two species,
were examined and the fusules definitely located : Agelenidae, Argiopidae, Dinopidae,
Filistatidae, Ochyroceratidae, Pliolcidae. Salticidae, Scytodidae, Tetragnathidae,
Theridiidae, Uloboridae.
A notable exception was the Amaurobiidae where the glands could not be located
though sections of two species of Ixeuticus and one of Matachia, and the skin of Amarara
were examined Nor were they found in sections of Bictyna, or of Ariadna, one of the
Segestriidae. Specimens belonging to the Thomisidae, Zoropsidae and Symphytognathidae
were examined superficially but fusules could not be detected.
The presence of the glands does not seem to be correlated with the habits of the spider.
They are well developed in the wandering, terrestrial GradunguZa, and in hole or tubeinhabiting forms such as the mygalomorphs and filistatids, but not in the amaurobiids.
They are also present in aerial web-spinners such as argiopids and uloborids. They
resemble the spinning glands so much in appearance that one wonders whether they and
not the true spinnerets are responsible for the sperm web. If so this would give support
to the suggestion that the median spiniierets are ventral glandular structures and are
not appendicular in nature.
REFERENCES
BEHLAND,
L., 1932. Les Arachnides. Encycl. e l i t . I’tcris. 16: 79-404.
BRISTOWE, W. S., 1930. Notes on t h e hiology of spiders 111. Ann. M a g . nrct. Hist. (10) 6: 347-353.
BRISTOWE,W. S. Ly- MILLOT,J., 1932. The Liphistiid Spiders. Proc. zool. Soc. Lond., 98: 1015-1057.
DAWYDOFF,
C., 1949. DQvelopment emhryonnaire des Arachnides. I n Trait6 de Zoologie. P . P . Grass&
Paris. 6 : 320-85.
FRITSCH,
A. 1904. Palaezoisclz Arachniden. Prague.
GERHARDT,
U. & KASTNEH,A., 1938. Araneae. I n Hanclbuch der Zoologie. Kukenthal, W. & Krumbach,
T. Berlin. 3ii: 394-656.
W. 1949. American Spiders. New York.
GEHTSCH,
H O L M A.
, 1940. Studien iiber die Entwicklung and Entwicklungsbiologie der Spinner. Zool. Bidr. Upps.,
19: 1-214.
KASTON,
B. J., 1948. Spiders of Connecticut. Bull. 70. Ceol. & r a t . Hist. Survey. Hartford, Conn.
KISHIDA,
K., 1923. Hrptathela, a new genus of liphistiid spiders. Annotes zool.jap., 10: 235-242.
222
B. J. MUPLES, F.L.S.
KOCH,
L., 1871-89. Die Arachniden Australiens. Nurnberg.
MACEADO,A. DE B., 1944. Observations inbdites sur le colulus et les filibres de quelques AranBides,
accompagnbes de notes critique sur la morphologie comparQedes filibres. Archos Mus. Bocage., 15:
13-52.
MACHADO,A. DE B., 1951. Ochyroceratidae (Araneae) de L’Angola. Publpes cult. Co. Diam. Avlgola,
8: 9-87.
MARPLES, B. J., 1955. Spiders from Western Samoa. J . Linn. Soc., 42: 453-504.
MILLOT, J., 1930. Colulus et’fili6res non funtionelles chez les AranQides. C.7. hebd. Siaanc. Acad. Sci.
Paris, 190: 209-210.
MILLOT, J., 1932. See Bristowe, W. & Millot, J.
MILLOT,J., 1949. Ordre des Palpigrades. I n Traitd de Zoologie. P. P. Grass&.Paris. 6: 520-32.
MONTGOMERY,
T. H., 1909. On the spinnerets, cribellum, colulus, tracheae and lung books of Araneads.
PTOC.
Acad. nut. Sci. Philad., 61 : 299-320.
PETRUNKEVITCH,
A., 1933. An inquiry into the natural classification of spiders, based on a study of
their internal anatomy. Trans. Conn. Acad. Arts Sci., 31 : 299-389.
PETRUNKEVITCR,
A., 1942. A study of amber spiders. Trans. Conn. Acad. Arts Sci., 34: 1 1 9 4 6 4 .
PETRUNKEVITCH,
A., 1949. A study of Palaeozoic Arachnida. Truns. Conn.. Acad. ArtsSci., 37: 69-315.
PETRUNKEVITCH,
A., 1953. Palaeozoic and Mesozoic Arachnida of Europe. Ceol. Soc. of America. Mem.
53.
SAVORY,
T. H., 1928. The Biology of Spiders. London.
SIMON,E., 1892-97. Histoire naturelle des Araignkees. Paris.
STORBXER,
L., 1949. Classe des Trilobites. I n Trait6 de Zoologie. P. P. Grass& Paris. 6: 160-97.
TIEGS,0. W., 1947. The development, and affinities of the Pauropoda. 11.&. JZ. microsc. Sci., 8 8 ; 256334.
VANDEL,A., 1949. Composition de l’embranchement. GencralitBs sur les Arthropods. I n Truitd de
Zoologie. P. P. Grass& Paris. 6 : 79-15s.
YOSHIKURA,
M., 1954. On the tracheae in a liphistiid spider, Heptathela kimurui. Kumamoto J . Sci. B.,
1: 37-40.
YOSHIKURA,
M., 1955. Embryological studies on the liphistiid spider Heptathela kimurai. Kumamoto
J . Sci. B., 2: 1-86.
YOSHIKURA,
M., 1958. On the development of a purse-web spider, Atypus kurschi Donitz. Kuntumoto
J . Sci. B.. 3 : 73-86.
EXPLANATION O F PLATE
PLATE
1
Fig. 1. Qr~du?7guZ~.
Parasagittal section of tho abdomen, anterior end to the right. Nine or ten of the
large epiandrous glands may be seen towards the lower right, between the genital furrow and the
waist. Scale, ca. 1.0 mm.
Fig. 2. Araneuspustulosus. Cleared cuticle of the ventral side of the abdomen. The curved transverse
line is the genital furrow. Anterior to it are ten pits of various sizes in which arise the fusules of the
epiandrous glands. Scale, ca. 0.26 mm.
Fig. 3. Gradungula. Part of a frontal section showing some of the epiandrous glands. On the left
their ducts traverse a tall glandular epithelium and open through fusules along the anterior side of
the genital furrow. Compare with 1. Scale, ca. 0.25 mm.
Fig. 4. Hickmania. Part of a sagittal section of the abdomen, anterior end to the left. Compare wlth
Text-fig. 3. The sperm duct, parked with sperms, is seen opening into the genital furrow. Between it
and the ventral surface are the epiandrous glands. TWOof their fusules are visible, and the region of
unicellular glands through which the ducts run. Scale, ca. 1.0 mm.
Note added in proof: Since writ)ingthis, my attention has been drawn by D. H. W. Levi
to it paper, “Zur Biologie der Vogelspinnen” by M. Melchers, 2. Morph. Okol. Tiere, 1964;
53: 517-536, in which are described the the structure and functions of the epiandrous
glands of Pamphobetus, a mygalomorph spider belonging to the family Theraphosidae. The
male spider spins, using its ordinary spinner, a sperm web in the form o f a sheet. Then,
using the epiandrous glands, it adds to this, from below, a small white sheet on which the
drop of sperm is placed. The spinning function of the epiandrous glands has thus actually
been observed.
J . Linn. IS". (Zool.) Vol. 46, N o . 310
Plate 1