As we stated in our brief general description of the visible hemisphere
of the moon, and as a cursory glance at our map and plates will have
shown, the predominant features of the lunar surface are the circular or
amphitheatrical formations that, by their number, and from their almost
unnatural uniformity of design, induced the belief among early observers
that they must have been of artificial origin. In proceeding now to
examine the details of our subject with more minuteness than before,
these annular formations claim the first share of our attention.

By general acceptation the term “crater” has been used to represent
nearly all the circular hollows that we observe upon the moon; and
without doubt the word in its literal sense, as indicating a _cup_ or
circular cavity, is so far aptly applied. But among geologists it has
been employed in a more special sense to define the hollowing out that
is found at the summit of some extinct, and the majority of active,
volcanoes. In this special sense it may be used by the student of the
lunar surface, though in some, and indeed in the majority of cases, the
lunar crater differs materially in its form with respect to its
surroundings from those on the earth; for while, as we have said, the
terrestrial crater is generally a hollow on a mountain top with its flat
bottom high above the level of the surrounding country, those upon the
moon have their lowest points depressed more or less deeply below the
general surface of the moon, the external height being frequently only a
half or one-third of the internal depth. Yet are the lunar craters truly
volcanic; as Sir John Herschel has said, they offer the true volcanic
character _in its highest perfection_. We have upon the earth some few
instances in which the geological conditions which have determined the
surface-formation have been identical with those that have obtained upon
the moon; and as a result we have some terrestrial volcanic districts
that, could we view them under the same circumstances, would be
identical in character with what we see by telescopic aid upon our
satellite. The most remarkable case of this similarity is offered by a
certain tract of the volcanic area about Naples, known from classic
times as the _Campi Phlegræi_, or burning fields, a name given to them
in early days, either because they showed traces of ancient earth-fire,
or because there were attached to the localities traditions concerning
hot-springs and sulphurous exhalations, if not of actual fiery
eruptions. The resemblance of which we are speaking is here so close
that Professor Phillips, in his work on Vesuvius, which by the way
contains a historical description of the district in question, calls the
moon a grand Phlegreian field. How closely the ancient craters of this
famous spot resemble the generality of those upon the moon may be judged
from Plate VI., in which representations of two areas, terrestrial and
lunar, of the same extent, are exhibited side by side, the terrestrial
region being the volcanic neighbourhood of Naples, and the lunar a
portion of the surface about the crater Theophilus.

In comparing these volcanic circles together, we are however brought
face to face with a striking difference that exists between the lunar
and terrestrial craters. This is the difference of magnitude. None of
those Plutonian amphitheatres included in the terrestrial area depicted
exceed a mile in diameter, and few larger volcanic vents than these are
known upon the earth. Yet when we turn to the moon, and measure some of
the larger craters there, we are astonished to find them ranging from an
almost invisible minuteness to 74 miles in diameter. The same
disproportion exists between the depths of the two classes of craters.
To give an idea of relative dimensions, we would refer to our
illustration of Copernicus[8] and its hundreds of comparatively minute
surrounding craters. Our terrestrial Vesuvius would be represented by
one of these last, which upon the plate measures about the twentieth of
an inch in diameter! And this disproportion strikes us the more forcibly
when we consider that the lunar globe has an area only one-thirteenth of
that of the earth. In view of this great apparent discrepancy it is not
surprising that many should have been incredulous as to the true
volcanic character of the lunar mountains, and have preferred to
designate them by some “non-committal” term, as an American geologist
(Professor Dana) has expressed it. But there is a feature in the
majority of the ring-mountains that, as we conceive, demonstrates
completely the fact of volcanic force having been in full action, and
that seems to stamp the volcanic character upon the crater-forms. This
special feature is the central cone, so well known as a characteristic
of terrestrial volcanoes, accepted as the result of the last expiring
effort of the eruptive force, and formed by the deposit, immediately
around the volcanic orifice, of matter which there was not force enough
to project to a greater distance. Upon the moon we have the central cone
in small craters comparable to those on the earth, and we have it in
progressively larger examples, upon all scales, up to craters of 74
miles in diameter, as we have shown in Plate VII. Where, then, can we
draw the line? Where can we say the parallel action to that which placed
Vesuvius in or near the centre of the arc of Somma, or the cone figured
in our sectional drawing of Vesuvius (Fig. 3) in the middle of its
present crater—where can we say that the action in question ceased to
manifest itself on the moon, seeing that there is no break in the
continuity of the crater-and-cone system upon the moon anywhere between
craters of 1¾ miles and 74 miles in diameter? We have, it is true, many
examples of coneless craters, but these are of all sizes, down to the
smallest, and up to a largeness that _would_ almost seem to render
untenable the ejective explanation: of these we shall specially speak in
turn, but for the present we will confine ourselves to the normal class
of lunar craters, those that have central cones, and that are in all
reasonable probability truly volcanic.

[Illustration: Fig. 16.]

And in the first place let us take a passing glance at the probable
formative process of a terrestrial volcano. Rejecting the hypothesis of
Von Buch, which geologists have on the whole found to be untenable, and
which ascribes the formation of all mountains to the elevation of the
earth’s crust by some thrusting power beneath, we are led to regard a
volcano as a pyramid of ejected matter, thrown out of and around an
orifice in the external solid shell of the earth by commotions
engendered in its molten nucleus. What is the precise nature and source
of the ejective force geologists have not perfectly agreed upon, but we
may conceive that highly expanded vapour, in all probability steam, is
its primary cause. The escaping aperture may have been a weak place
since the foundations of the earth were laid, or it may have been formed
by a local expansion of the nucleus in the act of cooling, upon the
principle enunciated in our Third Chapter; or, again, the expansile
vapour may have forced its own way through that point of the confining
shell that offered it the least resistance. The vent once formed, the
building of the volcanic mountain commenced by the out-belching of the
lava, ashes, and scoria, and the dispersion of these around the vent at
distances depending upon the energy with which they were projected. As
the action continued, the ejected matter would accumulate in the form of
a mound, through the centre of which communication would be maintained
with the source of the ejected materials and the seat of the explosive
agency. The height to which the pile would rise must depend upon several
conditions: upon the steady sustenance of the matter, and upon the form
and weight of the component masses, which will determine the slope of
the mountain’s sides. Supposing the action to subside gradually, the
tapering form will be continued upwards by the comparatively gentle
deposition of material around the orifice, and a perfect cone will
result of some such form as that represented below, which is the outline
ascribed by Professor Phillips to Vesuvius in pre-historic, or even
pre-traditional times, and which may be seen in its full integrity in
the cases of Etna, Teneriffe, Fussi-Yamma, the great volcanic mountain
of Japan, and many others. The earliest recorded form of Vesuvius is
that of a truncated cone represented in Fig. 17, which shows its
condition, according to Strabo, in the century preceding the Christian

[Illustration: PLATE VII

[Illustration: Fig. 17.]

[Illustration: Fig. 18.]

Now this form may have been assumed under two conditions. If, as
Phillips has surmised, the mountain originally had a peaked summit with
but a small crater-orifice at the point, then we must ascribe its
decapitation to a subsequent eruption which in its violence carried away
the upper portion, either suddenly, or through a comparatively slow
process of grinding away or widening out of the sides of the orifice by
the chafing or fluxing action of the out-going materials. But it is
probable that the mountain never had the perfect summit indicated in our
first outline. The violent outburst that caused the great crater-opening
of our second figure may have been but one paroxysmal phase of the
eruption that built the mountain: a sudden cessation of the eruptive
force when at its greatest intensity, and when the orifice was at its
widest, would leave matters in an opposite condition to that suggested
as the result of a slow dying out of the action: instead of the peak we
should have a wide crater-mouth. It is of small consequence for our
present purpose whether the crater was contemporaneous with the
primitive formation of the mountain, or whether it was formed centuries
afterwards by the blowing away of the mountain’s head; for upon the vast
scale of geological time, intervals such as those between successive
paroxysms of the same eruption, and those between successive eruptions,
are scarcely to be discriminated, even though the first be days and the
second centuries. We may remark that the widening of a crater by a
subsequent and probably more powerful eruption than that which
originally produced it is well established. We have only to glance at
the sketch, Fig. 18, of the outline of Vesuvius as it appeared between
the years A.D. 79 and 1631 to see how the old crater was enlarged by the
terrible Pompeian eruption of the first-mentioned year. Here we have a
crater ground and blown away till its original diameter of a mile and
three-quarters has been increased to nearly three miles. Scrope had no
hesitation in expressing his conviction that the external rings, such as
those of Santorin, St. Jago, St Helena, the Cirque of Teneriffe, the
Curral of Madeira, the cliff range that surrounds the island of Bourbon,
and others of similar form and structure, however wide the area they
enclose, are truly the “basal wrecks” of volcanic mountains that have
been blown into the air each by some eruption of peculiar paroxysmal
violence and persistence; and that the circular or elliptical basins
which they wholly or in part surround are in all cases true craters of

When the violent outburst that produces a great crater in a volcanic
mountain-top more or less completely subsides, the funnel or escaping
orifice becomes choked with débris. Still the vent strives to keep
itself open, and now and then gives out a small delivery of cindery
matter, which, being piled around the vent, after the manner of its
great prototype, forms the inner cone. This last may in its turn bear an
open crater upon its summit, and a still smaller cone may form within
_it_. As the action further dies away, the molten lava, no longer
seething and boiling, and spirting forth with the rest of the ejected
matter, wells upwards slowly, and cooling rapidly as it comes in contact
with the atmosphere, solidifies and forms a flat bottom or floor to the

[Illustration: Fig. 19.]

It may happen that a subsequent eruption from the original vent will be
comparable in violence to the original one, and then the inner cone
assumes a magnitude that renders it the principal feature of the
mountain, and reduces the old crater to a secondary object. This has
been the case with Vesuvius. During the eruption of 1631 the great cone
which we now call Vesuvius was thrown up, and the ancient crater now
distinguished as Monte Somma became a subsidiary portion of the whole
mountain. Then the appearance was that shown in Fig. 19, and which does
not differ greatly from that presented in the present day. The summit of
the Vesuvian cone, however, has been variously altered; it has been
blown away, leaving a large crateral hollow, and it has rebuilt itself
nearly upon its former model.

When we transfer our attention to the volcanoes of the moon, we find
ourselves not quite so well favoured with means for studying the process
of their formation; for the sight of the building up of a volcanic
mountain such as man has been permitted to behold upon the earth has not
been allowed to an observer of the moon. The volcanic activity,
enfeebled though it now be, of which we are witnesses from time to time
on the earth, has altogether ceased upon our satellite, and left us only
its effects as a clue to the means by which they were produced. If we in
our time could have seen the actual throwing up of a lunar crater, our
task of description would have been simple; as it is we are compelled to
infer the constructive action from scrutiny of the finished structure.

We can scarcely doubt that where a lunar crater bears general
resemblance to a terrestrial crater, the process of formation has been
nearly the same in the one case as in the other. Where variations
present themselves they may reasonably be ascribed to the difference of
conditions pertaining to the two spheres. The greatest dissimilarity is
in the point of dimensions; the projection of materials to 20 or more
miles distance from a volcanic vent appears almost incredible, until we
realize the full effect of the conditions which upon the moon are so
favourable to the dispersive action of an eruptive force. In the first
place, the force of gravity upon our satellite is only one-sixth of that
to which bodies are subject upon the earth. Secondly, by reason of the
small magnitude of the moon and its proportionally much larger surface
in ratio to its magnitude, the rate at which it parted with its cosmical
heat must have been much more rapid than in the case of the earth,
especially when enhanced by the absence of the heat-conserving power of
an atmosphere of air or water vapour; and the disruptive and eruptive
action and energy may be assumed to be greater in proportion to the more
rapid rate of cooling; operating, too, as eruptive action would on
matter so much reduced in weight as it is on the surface of the moon, we
thus find in combination conditions most favourable to the display of
volcanic action in the highest degree of violence. Moreover, as the
ejected material in its passage from the centre of discharge had not to
encounter any atmospheric resistance, it was left free to continue the
primary impulse of its ejection without other than gravitative
diminution, and thus to deposit itself at distances from its source
vastly greater than those of which we have examples on the earth.

We can of course only conjecture the source or nature of the moon’s
volcanic force. If geologists have had difficulty in assigning an origin
to the power that threw up our earthly volcanoes, into whose craters
they can penetrate, whose processes they can watch, and whose material
they can analyze, how vastly more difficult must be the inquiry into the
primary source of the power that has been at work upon the moon, which
cannot be virtually approached by the eye within a distance of six or
eight hundred miles, and the material of which we cannot handle to see
if it be compacted by heat, or distended by vapours. Steam is the agent
to which geologists have been accustomed to look for explanation of
terrestrial volcanoes; the contact of water with the molten nucleus of
our globe is accepted as a probable means whereby volcanic commotions
are set up and ejective action is generated. But we are debarred from
referring to steam as an element of lunar geology, by reason of the
absence of water from the lunar globe. We might suppose that a small
proportion of water once existed; but a small proportion would not
account for the immense display of volcanic action which the whole
surface exhibits. If we admitted a Neptunian origin to the disturbances
of the moon’s crust, we should be compelled to suppose that water had
existed nearly in as great quantity, area for area, there as upon our
globe; but this we cannot reasonably do.

[Illustration: PLATE VIII.

Aqueous vapour being denied us, we must look in other directions for an
ejective force. Of the nature of the lunar materials we can know
nothing, and we might therefore assume anything; some have had recourse
to the supposition of expansive vapours given off by some volatile
component of the said material while in a state of fusion, or generated
by chemical combinations. Professor Dana refers to sulphur as probably
an important element in the moon’s geology, suggesting this substance
because of the part which it appears to play in the volcanic or igneous
operations of our globe, and on account of its presence in cosmical
meteors that have come within range of our analysis. Any matter
sublimated by heat in the substrata of the moon would be condensed upon
reaching the cold surrounding space, and would be deposited in a state
of fine powder, or otherwise in a solid form. Maedler has attributed the
highly reflective portions of some parts of the surface, such as the
bright streams that radiate from some of the craters, Copernicus and
Tycho for instance, to the vitrification of the surface matter by
gaseous currents. But in suppositions like these we must remember that
the probability of truth diminishes as the free ground for speculation
widens. It does not appear clear how expansive vapours could have lain
dormant till the moon assumed a solid crust, as all such would doubtless
make their escape before any shell was formed, and at an epoch when
there was ample facility for their expansion.

While we are not insensible of the value of an expansive vapour
explanation, if it could be based on anything beyond mere conjecture, we
are disposed to attach greater weight to that afforded by the principle
sketched in our third chapter, viz., of expansion upon solidification.
We gave, as we think, ample proof that molten matter of volcanic nature,
when about passing to the solid state, increases its bulk to a
considerable degree, and we suggested that the lunar globe at one period
of its history must have been, what our earth is now, a solid shell
encompassing a molten nucleus; and further, that this last, in
approaching its solid condition, expanded and burst open or rent its
confining crust. At first sight it may seem that we are ascribing too
great a degree of energy to the expansive force which molten substances
exhibit in passing to the solid condition, seeing that in general such
forces are slow and gradual in their action; but this anomaly disappears
when we consider the vast bulk of the so expanding matter, and the
comparatively small amount that in its expansion it had to displace. It
is true that there are individual mountains on the moon covering many
square miles of surface, that as much as a thousand cubic miles of
material may have been thrown up at a single eruption; but what is this
compared to the entire bulk of the moon itself? A grain of mustard-seed
upon a globe three feet in diameter represents the scale of the loftiest
of terrestrial mountains; a similar grain upon a globe one foot in
diameter, would indicate the proportion of the largest upon the moon. A
model of our satellite with the elevations to scale would show nothing
more than a little roughness, or superficial blistering. Turn for a
moment to our map (Plate IV.), upon which the shadows give information
as to the heights of the various irregularities, and suppose it to
represent the actual size of some sphere whose surface has been broken
up by reactions of some kind of the interior upon the exterior—suppose
it to have been a globe of fragile material filled with some viscous
substance, and that this has expanded, cracked its shell, oozed out in
the process of solidification, and solidified: the irregularity of
surface which the small sphere, roughened by the out-leaking matter,
would present, would not be less than that exhibited in the map under
notice. When we say that a lunar crater has a diameter of 30 miles, we
raise astonishment that such a structure could result from an eruption
by the expansive force of solidifying matter; but when we reflect that
this diameter is less than the two-hundredth part of the circumference
of the moon, we need have no difficulty in regarding the upheaval as the
result of a force slight in comparison to the bulk of the material
giving rise to it. We have upon the moon evidence of volcanic eruptions
being the final result of most extensive dislocations of surface, such
as could only be produced by some widely diffused uplifting force. We
allude to the frequent occurrence of chains of craters lying in a nearly
straight line, and of craters situated at the converging point of
visible lines of surface disturbance. Our map will exhibit many examples
of both cases. An examination of the upper portion (the southern
hemisphere of the moon) will reveal abundant instances of the linear
arrangement, three, four, five or even more crateral circles will be
found to lie with their centres upon the same great-circle track,
proving almost undoubtedly a connexion between them so far as the
original disturbing force which produced them is concerned. Again, in
the craters Tycho (30), Copernicus (147), Kepler (146), and Proclus
(162), we see instances of the situation of a volcanic outburst at an
obvious focus of disturbance. These manifest an up-thrusting force
covering a large sub-surface area, and escaping at the point of least
resistance. Such an extent of action almost precludes the gaseous
explanation, but it is compatible with the expansion on consolidation
theory, since it is reasonable to suppose that in the process of
consolidation the viscous nucleus would manifest its increase of bulk
over considerable areas, disturbing the superimposed crust either in one
long crack, out of the wider opening parts of which the expanded
material would find its escape, or “starring” it with numerous cracks,
from the converging point of which the confined matter would be ejected
in greatest abundance and, if ejected there with great energy and
violence, would result in the formation of a volcanic crater.

The actual process by which a lunar crater would be formed would differ
from that pertaining to a terrestrial crater only to the extent of the
different conditions of the two globes. We can scarcely accept Scrope’s
term “basal wrecks” (of volcanic mountains that have had the summits
blown away) as applicable to the craters of the moon, for the reason
that the lunar globe does not offer us any instance of a mountain
comparable in extent to the great craters and whose summit has _not_
been blown away. Scrope’s definition implies a double, or divided
process of formation: first the building up of a vast conical hill and
then the decapitation and “evisceration” of it at some later period.
There are grounds for this inferred double action among the terrestrial
volcanoes, since both the perfect cone and its summitless counterpart
are numerously exemplified. But upon the moon we have no perfect cone of
great size, we have no exception whereby the rule can be proved. It is
against probability, supposing every lunar crater to have once been a
mountain, that in every case the mountain’s summit should have been
blown away; and we are therefore compelled to consider that the moon’s
volcanic craters were formed by one continuous outburst, and that their
“evisceration” was a part of the original formative process. We do not,
however, include the central cone in this consideration: that may be
reasonably ascribed to a secondary action or perhaps, better, to a
weaker or modified phase of the original and only eruption.

[Illustration: Fig. 20.]

[Illustration: Fig. 21.]

[Illustration: PLATE IX.

Under these circumstances we conceive the upcasting and excavating of a
normal lunar crater to have been primarily caused by a local
manifestation of the force of expansion upon solidification of the
subsurface matter of the moon, resulting in the creation of a mere
“star” or crack in and through the outermost and solid crust. As we
shall have to rely upon diagrams to explain the more complicated
features, we give one of this elementary stage also as a commencement of
the series; and Fig. 20 therefore represents a probable section of the
lunar surface at a point which was subsequently the location of a
crater. From the vent thus formed we conceive the pent-up matter to have
found its escape, not necessarily at a single outburst, but in all
probability in a paroxysmal manner, as volcanic action manifests itself
on our globe. The first outflow of molten material would probably
produce no more than a mere hill or tumescence as shewn sectionally in
Fig. 21; and if the ejective force were small this might increase to the
magnitude of a mountain by an exudative process to be alluded to
hereafter. But if the ejective force were violent, either at the moment
of the first outburst or at any subsequent paroxysm, an action
represented in Fig. 22 would result: the unsupported edges or lips of
the vent-hole would be blown and ground or fluxed away, and a
funnel-formed cavity would be produced, the ejected matter (so much of
it as in falling was not caught by the funnel) being deposited around
the hollow and forming an embryo circular mountain. The continuance of
this action would be accompanied by an enlargement of the conical cavity
or crater, not only by the outward rush of the violently discharged
material, but also by the “sweating” or grinding action of such of it as
in descending fell within the hollow. And at the same time that the
crater enlarged the rampart would extend its circumference, for it would
be formed of such material as did not fall back again into the crater.
Upon this view of the crater-forming process we base the sketch, Fig.
23, of the probable section of a lunar crater at one period of its

[Illustration: Fig. 22.]

So long as each succeeding paroxysm was greater than its predecessor,
this excavating of the hollow and widening of its mouth and mound would
be extended. But when a weaker outburst came, or when the energy of the
last eruption died away, a process of slow piling up of matter close
around the vent would ensue. It is obvious that when the ejective force
could no longer exert itself to a great distance it must merely have
lifted its burden to the relieving vent and dropped it in the immediate
neighbourhood. Even if the force were considerable, the effect, so long
as it was insufficient to throw the ejecta beyond the rim of the crater,
would be to pile material in the lowermost part of the cavity; for what
was not cast over the edge would roll or flow down the inner slope and
accumulate at the bottom. And as the eruption died away, it would add
little by little to the heap, each expiring effort leaving the out-given
matter nearer the orifice, and thus building up the central cone that is
so conspicuous a feature in terrestrial volcanoes, and which is also a
marked one in a very large proportion of the craters of the moon. This
formation of the cone is pictorially described by Fig. 24.

[Illustration: Fig. 23.]

[Illustration: Fig. 24.]

In the volcanoes of the earth we observe another action either
concurrent with or immediately subsequent to the erection or formation
of the cone: this is the outflow or the welling forth of fluid lava,
which in cooling forms the well-known plateau. We have this feature
copiously represented upon the moon and it is presumable that it has in
general been produced in a manner analogous to its counterparts upon the
earth. We may conceive that the fluid matter was either spirted forth
with the solid or semisolid constituents of the cone, in which case it
would drain down and fill the bottom of the crater; or we may suppose
that it issued from the summit of the cone and ran down its sides, or
that, as we see upon the earth, it found its escape before reaching the
apex, by forcing its way through the basal parts. These actions are
indicated hypothetically for the moon in Fig. 25; and the parallel
phenomena for the earth are shewn by the actual case (represented in
Fig. 26 and on Plate I.) of Vesuvius as it was seen by one of the
authors in 1865, when the principal cone was vomiting forth ashes,
stones, and red-hot lava, while a vent at the side emitted very fluid
lava which was settling down and forming the plateau.

[Illustration: Fig. 25.]

Although we cannot, obviously, see upon the moon evidence of a cone
actually overtopped by the rising lake of lava, yet it is not
unreasonable to suppose that such a condition of things actually
occurred in many of those instances in which we observe craters without
central cones, but with plateaux so smooth as to indicate previous
fluidity or viscosity. From the state of things exhibited in Fig. 25 the
transition to that shewn in Fig. 27 is easily, and to our view
reasonably, conceivable. We are in a manner led up to this idea by a
review of the various heights of central cones above their surrounding
plateaux. For instance, in such examples as Tycho or Theophilus, we have
cones high above the lava floor; in Copernicus, Arzachael and Alphonsus
they are comparatively lower; the lava in these and some other craters
does not appear to have risen so high; while in Aristotle and Eudoxus
among others, we have only traces of cones, and it is supposable that in
these cases the lava rose so high as nearly to overtop the central
cones. Why should it not have risen so far as to overtop and therefore
conceal some cones entirely? We offer this as at least a feasible
explanation of some coneless craters: it is not necessary to suppose
that it applies to all such, however: there may have been many craters,
the formation of which ceased so abruptly that no cone was produced,
though the welling forth of lava occurred from the vent, which may have
been left fully open, as in Fig. 28, or so far choked as to stay the
egress of solid ejecta and yet allow the fluid material to ooze upwards
through it, and so form a lake of molten lava which on consolidation
became the plateau. As most of the examples of coneless craters exhibit
on careful examination minute craters on the surface of the otherwise
smooth plateaux, we may suppose that such minute craters are evidences
of the upflow of lava which resulted in the plateaux.

[Illustration: PLATE X.

[Illustration: Fig. 26.]

[Illustration: Fig. 27.]

[Illustration: Fig. 28.]

We have strong evidence in support of this up-flow of lava offered by
the case of the crater Wargentin, (No. 26, 57·5—140·2) situated near the
south-east border of the disc, and of which we give a special plate.
(Plate XVII.) It appears to be really a crater in which the lava has
risen almost to the point of overflowing, for the plateau is nearly
level with the edge of the rampart. This edge appears to have been
higher on one side than the other, for on the portion nearest the centre
of the visible disc we may, under favourable circumstances, detect a
segment of the basin’s brim rising above the smooth plateau as indicated
in our illustration. Upon the opposite side there is no such feature
visible, the plateau forms a sharp table-like edge. It is just possible
that an actual overflow of lava took place at this part of the crater,
but from the unfavourable situation of this remarkable object it is
impossible to decide the point by observation. There is no other crater
upon the visible hemisphere of the moon that exhibits this filled-up
condition; but, unique as it is, it is sufficient to justify our
conclusion that the plateau-forming action upon the moon has been a
flowing-up of fluid matter from below subsequent to the formation of the
crater-rampart, and not, as a casual glance at the great smooth-bottom
craters might lead us to suspect, a result of some sort of diluvial
deposit which has filled hollows and cavities and so brought up an even
surface. The elevated basin of Wargentin could not have been filled thus
while the surrounding craters with ramparts equally or less high
remained empty: its contained matter must have been supplied from
within, we must conjecture by the upflow of lava from the orifice which
gave forth the material to form the crateral rampart in the first
instance. We are free to conjecture that at some period of this
table-mountain’s formation it was a crater with a central cone, and that
the rising lava over-topped this last feature in the manner shewn by the
above figure (Fig. 29).

[Illustration: Fig. 29.]

The question occurs whether other craters may not have been similarly
filled and have emptied themselves by the bursting of the wall under the
pressure of the accumulated lake of lava within. We know that this
breaching is a common phenomenon in the volcanoes of our globe; the
district of Auvergne furnishing us with many examples; and there are
some suspicious instances upon the moon. Copernicus exhibits signs of
such disruption, as also does the smaller crater intruding upon the
great circle of Gassendi. (See Frontispiece.) But the existence of such
discharging breaches implies the outpouring of a body of fluid or
semi-fluid material, comparable in cubical content to the capacity of
the crater, and of this we ought to see traces or evidence in the form
of a bulky or extensive lava stream issuing from the breach. But
although there are faint indications of once viscous material lying in
the direction that escaping fluid would take, we do not find anything of
the extent that we should expect from the mass of matter that would
constitute _a craterfull_. It is true that if the escaping fluid had
been very limpid it might have spread over a large area and have formed
a stratum too thin to be detected. Such a degree of limpidity as would
be required to fulfil this condition we are hardly, however, justified
in assuming.

To return to the subject of central cones. Although there are cases in
which the simple condition of a single cone exists, yet in the majority
we see that the cone-forming process has been divided or interrupted,
the consequence being the production of a group of conical hills instead
of a single one. Copernicus offers an example of this character, six,
some observers say seven, separate points of light, indicating as many
peaks tipped with sunshine, having been seen when the greater part of
the crater has been buried in shadow. Erastothenes, Bulialdus,
Maurolicus, Petavius, Langreen, and Gassendi, are a few among many
instances of craters possessing more than a central single cone. This
multiplication of peaks upon the moon doubtless arose from similar
causes to those which produce the same feature in terrestrial volcanoes.
Our sketch of Vesuvius in 1865 (Fig. 26) shows the double cone and the
probable source of the secondary one in the diverted channel of the
out-coming material. A very slight interruption in the first instance
would suffice to divert the stream and form another centre of action, or
a choking of the original vent would compel the issuing matter to find a
less resisting thoroughfare into open space, and the process of
cone-building would be continued from the new orifice, perhaps to be
again interrupted after a time and again driven in another direction. In
this manner, by repeated arrests and diversions of the ejecta, cone has
grown upon the side of cone, till, ere the force has entirely spent
itself, a cluster of peaks has been produced. It may have been that this
action has taken place after the formation of the plateau, in the manner
indicated by Fig. 30; a spasmodic outburst of comparatively slight
violence having sought relief from the original vent, and the flowing
matter, finding the one orifice not sufficiently open to let it pass,
having forced other exit through the plateau.

[Illustration: PLATE XI.

[Illustration: Fig. 30.]

In frequent instances we observe the state of things represented in Fig.
31, in which the plateau is studded with few or many small craters. This
is the case with Plato, with Arzachael, Hipparchus, Clavius (which
contains about 15 small internal craters), and many others. It is
probable that these subsidiary craters were produced by an after-action
like that which has produced duplicated cones, but in which the
secondary eruption has been of somewhat violent character, for it may
almost be regarded as an axiom that violent eruptions excavate craters
and weak ones pile up cones. In the cases referred to it seems
reasonable to suppose that the main vent has been the channel for an
up-cast of material, but that at some depth below the surface this
material met with some obstruction or cause of diversion, and that it
took a course which brought it out far away from the cone upon the floor
of the plateau. It might even be carried so far as to be upon the
rampart, and it is no uncommon thing to see small craters in such a
situation, though when they appear at such a distance from the primary
vent, it seems more reasonable to suppose that they do not belong to it
but have arisen from a subsequent and an independent action.

[Illustration: Fig. 31.]

We find scarcely an instance of a small crater occurring just in the
centre of a large one, or taking the place of the cone. This is a
curious circumstance. Whenever we have any central feature in a great
crater, that feature is a cone. The tendency of this fact is to prove
that cones were produced by very weak efforts of this expiring force,
for had there been any strength in the last paroxysm it is presumable
that it would have blown out and left a crater. No very violent
eruptions have therefore taken place from the vents that were connected
with the great craters of the moon, nothing more powerful than could
produce a cone of exudation or a cinder-heap. And with regard to cones,
it is noteworthy that whether they be single or multiple, they never
rise so high as the circular ramparts of their respective craters. This
supports the inferred connexion between the crater origin and the cone
origin, for supposing the two to have been independent, a supposition
untenable in view of the universality of the central position of the
cone, it is scarcely conceivable that the mountains should have always
been located within ramparts higher than themselves. The less height
argues less power in the upcasting agency, and the diminished force may
well be considered as that which would almost of necessity precede the
expiration of the eruption.

Occasionally a crater is met with that has a double rampart, and the
concentricity suggests that there have been two eruptions from the same
vent: one powerful, which formed the exterior circle, and a second
rather less powerful which has formed the interior circle. It is not,
however evident that this duplication of the ring has always been due to
a double eruption. In many cases there is duplication of only a portion:
a terrace exhibits itself around a part of the circular range, sometimes
upon the outside and sometimes upon the inside. These terraces are not
likely to have been formed by any freak of the eruption, and we are led
to ascribe them in general to landslip phenomena. When, in the course of
a volcano’s formation, the piling-up of material about the vent has
continued till the lower portions have been unable to support the upper,
or when from any cause, the material composing the pile has lost its
cohesiveness, the natural consequence has been a breaking away of a
portion of the structure and its precipitation down the inclined sides
of the crater. Vast segments of many of the lunar mountain-rings appear
to have been thus dislodged from their original sites and cast down the
flanks to form crescent ranges of volcanic rocks either within or
without the circle. Nearly every one of our plates contains craters
exhibiting this feature in more or less extensive degree. Sometimes the
separated portion has been very small in proportion to the circumference
of the crater: Plato is an instance in which a comparatively small mass
has been detached. In other cases very large segments have slid down and
lie in segmental masses on the plateaux or form terraces around the
rampart. Aristarchus, Treisnecker and Copernicus exhibit this larger
extent of dislocation.

It is possible that these landslips occurred long after the formation of
the craters that have been subject to them. They are probably
attributable to recent disintegration of the lunar rocks, and we have a
powerful cause for this in the alternations of temperature to which the
lunar crust is exposed. We shall have occasion to revert to this subject
by-and-bye; at present it must suffice to point out that the extremes of
cold and heat, between which the lunar soil varies, are, with reasonable
probability, assumed to be on the one hand the temperature of space
(which is supposed to be about 200° below zero), and, on the other hand,
a degree of heat equal to about twice that of boiling water. A range of
at least 500° must work great changes in such heterogeneous materials as
we may conjecture those of the lunar crust to be, by the alternate
contractions and expansions which it must engender, and which must tend
to enlarge existing fissures and create new ones, to grind contiguous
surfaces and to dislodge unstable masses. This cause of change, it is to
be remarked, is one which is still exerting itself.

In a few cases we have an entirely opposite interruption of the
uniformity of a crater’s contour. Instead of the breaking away of the
ring in segments we see the entire circuit marked with deep ruts that
run down the flanks in a radial direction, giving us evidence of a
downward _streaming_ of semi-fluid matter, instead of a disruption of
solid masses. We cannot doubt that these ruts have been formed by lava
currents, and they indicate a condition of ejected material different
from that which existed in the cases where the landslip character is
found. In these last the ejecta appears to have been in the form of
masses of solidified or rapidly solidifying matter, which remained where
deposited for a time and then gave way from overloading or loss of
cohesiveness, whereas the substances thrown out in the case of the
rutted banks were probably mixed solid and fluid, the former remaining
upon the flanks while the latter trickled away. Nothing so well
represents, upon a small scale, this radial channelling as a heap of
wetted sand left for a while for the water to drain off from it. The
solid grains in such a heap sustain its general mass-form, but the
liquid in passing away cuts the surface into fissures running from the
summit to the base, and forms it into a model of a volcanic mountain
with every feature of peak, crag, and chasm reproduced, This similarity
of effect leads us to suspect a parallelism of cause, and thus to the
inference that the material which originally formed such a
crater-mountain as Aristillus (which is a most prominent example of this
rutted character, and appears in Plate IX., side by side with a crater
that has its banks segmentally broken), must have been of the compound
nature indicated; and that an action analogous to that which ruts a damp
sand-heap, rutted also the banks of the lunar crater.

[Illustration: PLATE XII.

Before passing from the subject of craters it behoves us to say a few
words upon the curious manner in which these formations are complicated
by intermingling and superposition. Yet, upon this point, we may be
brief, for in the way of description our plates speak more forcibly than
is possible by words. In particular we would refer to Plate XII., which
represents the conspicuous group of craters of which the three largest
members have been respectively named Theophilus, Cyrillus, and
Catherina. But the area included in this plate is by no means an
extraordinary one; there are regions about Tycho wherein the craters so
crowd and elbow each other that, in their intricate combinations, they
almost defy accurate depiction. Our map and Plate XVI. will serve to
give some idea of them. This intermingling of craters obviously shows
that all the lunar volcanoes were not simultaneously produced, but that
after one had been formed, an eruption occurred in its immediate
neighbourhood and blew a portion of it away; or it may have been that
the same deep-seated vent at different times gave forth discharges of
material the courses of which were more or less diverted on their way to
the surface.

We have before alluded to the frequent occurrence of lines of craters
upon the moon. In these lines the overlapping is frequently visible; it
is seen in Plate XII. before referred to, where the ring mountains are
linked into a chain slightly curved, and upon the map, Plate IV., the
nearly central craters Ptolemy and Alphonsus, the latter of which
overlaps the former, are seen to form part of a line of craters marking
a connection of primary disturbance. An extensive crack suggests itself
as a favourable cause for the production of this overlaying of craters,
for it would serve as a sort of “line of fire” from various points at
which eruptions would burst forth, sometimes weak or far apart, when the
result would be lines of isolated craters, and sometimes near together,
or powerful, when the consequence would be the intrusion of one upon the
other, and the perfect production of the latest formed at the expense or
to the detriment of those that had been formed previously. The linear
grouping of volcanoes upon the earth long ago struck observant minds.
The fable of the _Typhon_ lying under Sicily and the Phlegreian fields
and disturbing the earth by its writhings, is a mythological attempt to
explain the particular case in that region.

The capricious manner in which these intrusions occur is very curious.
Very commonly a small crater appears upon the very rampart of a greater
one, and a more diminutive one still will appear upon the rampart of the
parasite. Stoeffler presents us with one example of this character,
Hipparchus with another, Maurolycus with a third, and these are but a
few cases of many. Here and there we observe several craters ranged in a
line with their rims in one direction all perfect, and the whole
appearing like a row of coins that have fallen from a heap. There is an
example near to Tycho which we reproduce in Plate XX. In this case one
is led to conjecture that the ejective agency, after exerting itself in
one spot, travelled onward and renewed itself for a time; that it ceased
after forming crater number two, and again journeyed forward in the same
line, recommencing action some miles further, and again subsiding; yet
again pushing forward and repeating its outburst, till it produced the
fourth crater, when its power became expended. In each of these
successive eruptions the centre of discharge has been just outside the
crater last formed; and the close connexion of the members of the group,
together with the fact of their nearly similar size, appears to indicate
a community of origin. For it seems feasible that as a general rule the
size of a crater may be taken as a measure of the depth of force that
gave rise to the eruption producing it. This may not be true for
particular cases, but it will hold where a great number are collectively
considered; for if we assume the existence of an average disturbing
force, it is apparently clear that it will manifest itself in disturbing
greater or less surface-areas in proportion as it acts from greater or
less depths. Or, _mutatis mutandis_, if we assume an uniform depth for
the source of action, the greater or less surface disturbance will be a
measure of greater or less eruptive intensity.

Perhaps the most remarkable case of a vast number of craters, which,
from their uniform dimensions, suggest the idea of community of
source-power or source-depth, is that offered by the region surrounding
Copernicus, which, as will be seen by our plate of that object, is a
vast Phlegreian field of diminutive craters. So countless are the minute
craters that a high magnifying power brings into view when atmospheric
circumstances are favourable, and so closely are they crowded together,
that the resulting appearance suggests the idea of froth, and we should
be disposed to christen this the “frothy region” of the moon, did not a
danger exist in the tendency to connect a name with a cause. The craters
that are here so abundant are doubtless the remains of true volcanoes
analogous to the parasitical cones that are to be found on several
terrestrial mountains, and not such accidental formations as the
_Hornitos_ described by Humboldt as abounding in the neighbourhood of
the Mexican volcano, Jurillo, but which the traveller did not consider
to be true cones of eruption.[9] Although upon our plate, and in
comparison with the great crater that is its chief feature, these
countless hollows appear so small as at first sight to appear
insignificant, we must remember that the minutest of them must be grand
objects, each probably equal in dimensions to Vesuvius. For since, as we
have shown in an early chapter, the smallest discernible telescopic
object must subtend an angle to our eye of about a second, and since
this angle extended to the moon represents a mile of its surface, it
follows that these tiny specks of shadow that besprinkle our picture,
are in the reality craters of a mile diameter. This comparison may help
the conception of the stupendous magnitude of the moon’s volcanic
features; for it is a conception most difficult to realize. It is hard
to bring the mind to grasp the fact that that hollow of Copernicus is
fifty miles in diameter. We read of an army having encamped in the once
peaceful crater of Vesuvius, and of one of the extinct volcanoes of the
_Campi Phlegræi_ being used as a hunting preserve by an Italian king.
These facts give an idea of vastness to those who have not the good
fortune to see the actual dimensions of a volcanic orifice themselves.
But it is almost impossible to conjure up a vision of what that
fifty-mile crater would look like upon the moon itself; and for want of
a terrestrial object as a standard of comparison, our picture, and even
the telescopic view of the moon itself, fails to render the imagination
any help. We may try to realize the vastness by considering that one of
our average English counties could be contained within its ramparts, or
by conceiving a mountainous amphitheatre whose opposite sides are as far
apart as the cathedrals of London and Canterbury, but even these
comparisons leave us unimpressed with the true majesty which the object
would present to a spectator upon the surface of our satellite.

In our previous chapter we have given a reason for regarding as true
volcanic craters all those circular formations, of whatever size, that
exhibit that distinctive feature _the central cone_. Between the
smallest crater with a cone that we can detect under the best telescopic
conditions, namely, the companion to Hell, 1¾ mile diameter, and the
great one called Petavius, 78 miles in diameter, we find no break in the
continuity of the crater-cum-cone system that would justify us in saying
that on the one side the volcanic or eruptive cause ceased, and on the
other side some other causative action began. But there are numerous
circular formations that surpass the magnitude of Petavius and its
peers, but that have no central cone, and are, therefore, not so
manifestly volcanic as those which possess this feature. Our map will
show many striking examples of this class at a glance. We may in
particular refer _inter alia_ to Ptolemy near the centre of the moon, to
Grimaldi (No. 125), Shickard (No. 28), Schiller (No. 24), and Clavius
(No. 13), all of which exceed 100 miles in diameter. Even the great
_Mare Crisium_, nearly 300 miles in diameter, appears to be a formation
not distinct from those which we have just named. These present little
of the generic crater character in their appearance; and they have been
distinguished therefrom by the name of _Walled_ or _Ramparted Plains_.
Their actual origin is beyond our explanation, and in attempting to
account for them we must perforce allow considerable freedom to
conjecture. They certainly, as Hooke suggested, present a “broken
bubble”-like aspect; but one cannot reasonably imagine the existence of
any form of mineral matter that would sustain itself in bubble form over
areas of many hundreds of square miles. And if it were reasonable to
suppose the great rings to be the foundations of such vast volcanic
domes, we must conclude these to have broken when they could no longer
sustain themselves, and in that case the surface beneath should be
strewed with _débris_, of which, however, we can find no trace.
Moreover, we might fairly expect that some of the smaller domes would
have remained standing: we need hardly say that nothing of the kind

[Illustration: Fig. 32.]

The true circularity of these objects appears at first view a remarkable
feature. But it ceases to be so if we suppose them to have been produced
by some very concentrated sublunar force of an upheaving nature, and if
only we admit the homogeneity of the moon’s crust. For if the crust be
homogeneous, then _any_ upheaving force, deeply seated beneath it, will
exert itself _with equal effects at equal distances from the source_:
the lines of equal effect will obviously be radii of a sphere with the
source of the disturbance for its centre, and they will meet a surface
over the source in a circle. This will be evident from Fig. 32, in which
a force is supposed to act at F below the surface s s s s. The matter
composing s s being homogeneous, the action of F will be equal at equal
distances in all directions. The lines of equal force, F_ f_, F _f_,
will be of equal length, and they will form, so to speak, radii of a
sphere of force. This sphere is cut by the plane at _s s s s_, and as
the intersection necessarily takes place everywhere at the extremity of
these radii, the figure of intersection is demonstrably a circle (shown
in perspective as an ellipse in the figure). Thus we see that an intense
but extremely confined explosion, for instance, beneath the moon’s crust
must disturb a _circular_ area of its surface, if the intervening
material be homogeneous. If this be not homogeneous there would be,
where it offered _less_ than the average resistance to the disturbance,
an outward distortion of the circle; and an opposite interruption to
circularity if it offers _more_ than the average resistance. This
assumed homogeneity may possibly be the explanation of the general
circularity of the lunar surface features, small and great.

[Illustration: Fig. 33.]

[Illustration: Fig. 34.]

We confess to a difficulty in accounting for such a very local
generation of a deep-seated force; and, granting its occurrence, we are
unprepared with a satisfactory theory to explain the resultant effect of
such a force in producing a raised ring at the limit of the circular
disturbance. We may indeed, suppose that a vast circular cake or conical
frustra would be temporarily upraised as in Fig. 33, and that upon its
subsidence a certain extrusion of subsurface matter would occur around
the line or zone of rupture as in Fig. 34. This supposition, however,
implies such a peculiarly cohesive condition of the matter of the
uplifted cake, that it is doubtful whether it can be considered tenable.
We should expect any ordinary form of rocky matter subjected to such an
upheaval to be fractured and distorted, especially when the original
disturbing force is greater in the centre than at the edge, as,
according to the above hypothesis, it would be; and in subsiding, the
rocky plateau would thus retain some traces of its disturbance; but in
the circular areas upon the moon there is nothing to indicate that they
have been subjected to such dislocations.

[Illustration: Fig. 35. A A. Fissures gaping downwards and injected
by intumescent lava beneath. B B B. Fissures gaping upwards and
allowing wedges of rock to drop below the level of the intervening
masses, C C. Wedges forced upwards by horizontal compression. E F.
Neutral plane or pivot axis, above and below which the directions of
the tearing strain and horizontal compression are severally
indicated by the smaller arrows; the larger arrows beneath represent
the direction of the primary expansive force.]

Mr. Scrope in his work on volcanoes has given a hypothetical section of
a portion of the earth’s crust, which presents a bulging or tumescent
surface in some measure resembling the effect which such a cause as we
have been considering would produce. We give a slightly modified version
of his sketch in Fig. 35, showing what would be the probable phenomena
attending such an upheaval as regards the behaviour of the disturbed
portion of the crust, and also that of the lava or semifluid matter
beneath: and, as will be seen by the sketch, a possible phase of the
phenomena is the production of an elevated ridge or rampart at the
points of disruption _c c_; and where there is a ring of disruption, as
by our hypothesis there would be, the ridge or rampart _c c_ would be a
circle. In this drawing we see the cracking and distortion to which the
elevated area would be subjected, but of which, as previously remarked,
the circular areas of the moon present no trace of residual appearance.

[Illustration: PLATE XIV.

Those who have offered other explanations of these vast ring-formed
mountain ranges, have been no more happy in their conjectures. M. Rozet,
who communicated a paper on selenology to the French Academy in 1846,
put forth the following theory. He argued that during the formation of
the solid scoriaceous pelicules of the moon, circular or tourbillonic
movements were set up; and these, by throwing the scoria from the centre
to the circumference, caused an accumulation thereof at the limit of the
circulation. He considered that this phenomenon continued during the
whole process of solidification, but that the amplitude of the whirlpool
diminished with the decreasing fluidity of the surface material.
Further, he suggested that when many vortices were formed, and the
distances of their centres, taken two and two, were less than the sums
of their radii, there resulted closed spaces terminated by arcs of
circles; and when for any two centres the distance was greater than the
sum of the radii of action, two separate and complete rings were formed.
We have only to remark on this, that we are at a loss to account for the
origination of such vorticose movements, and M. Rozet is silent on the
point. If the great circles are to be referred to an original sea of
molten matter, it appears to us more feasible to consider that wherever
we see one of them there has been, at the centre of the ring, a great
outflow of lava that has flooded the surrounding surface. Then, if from
any cause, and it is not difficult to assign one, the outflow became
intermittent, or spasmodic, or subject to sudden impulses, concentric
waves would be propagated over the pool and would throw up the scoria or
the solidifying lava in a circular bank at the limit of the fluid area.

This hypothesis does not differ greatly from the _ebullition_ theory
proposed by Professor Dana, the American geologist, to explain these
formations. He considered that the lunar ring-mountains were formed by
an action analogous to that which is exemplified on the earth in the
crater of Kilauea, in the Hawaiian islands. This crater is a large open
pit exceeding three miles in its longer diameter, and nearly a thousand
feet deep. It has clear bluff walls round a greater part of its circuit,
with an inner ledge or plain at their base, raised 340 feet above the
bottom. This bottom is a plain of solid lavas, entirely open to day,
which may be traversed with safety (we are quoting Professor Dana’s own
statement written in 1846, and therefore not correctly applying to the
present time): over it there are pools of boiling lava in active
ebullition, and one is more than a thousand feet in diameter. There are
also cones at times, from a few yards to two or three thousand feet in
diameter, and varying greatly in angle of inclination. The largest of
these cones have a circular pit or crater at the summit. The great pit
itself is oblong, owing to its situation on a fissure, but the lakes
upon its bottom are round, and in them, says Professor Dana, “the
circular or slightly elliptical form of the moon’s craters is
exemplified to perfection.”

Now Dana refers this great pit crater and its contained lava-lakes to
“the fact that the action at Kilauea is simply _boiling_, owing to the
extreme fluidity of the lavas. The gases or vapours which produce the
state of active ebullition escape freely in small bubbles, with little
commotion, like jets over boiling water; while at Vesuvius and other
like cones they collect in immense bubbles before they accumulate force
enough to make their way through; and consequently the lavas in the
latter case are ejected with so much violence that they rise to a height
often of many thousand feet and fall around in cinders. This action
builds up the pointed mountain, while the simple boiling of Kilauea
makes no cinders and no cinder cones.”

Professor Dana continues, “If the fluidity of lavas, then, is sufficient
for this active ebullition, we may have boiling going on over an area of
an indefinite extent; for the size of a boiling lake can have no limits
except such as may arise from a deficiency of heat. The size of the
lunar craters is therefore no mystery. Neither is their circular form
difficult of explanation; for a boiling pool necessarily, by its own
action, extends itself circularly around its centre. The combination of
many circles, and the large sea-like areas are as readily

In justice to Professor Dana it should be stated that he included in
this theory of formation all lunar craters, even those of small size and
possessing central cones; and he put forth his views in opposition to
the eruptive theory which we have set forth, and which was briefly given
to the world more than twenty-five years ago. As regards the smallest
craters with cones, we believe few geologists will refuse their
compliance with the supposition that they were formed as our
crater-bearing volcanoes were formed: and we have pointed out the
logical impossibility of assigning any limit of size beyond which the
eruptive action could not be said to hold good, so long as the central
cone is present. But when we come to ring-mountains having no cones, and
of such enormous size that we are compelled to hesitate in ascribing
them to ejective action, we are obliged to face the possibility of some
other causation. And, failing an explanation of our own that satisfied
us, we have alluded to the few hypotheses proffered by others, and of
these Professor Dana’s appears the most rational, since it is based upon
a parallel found on the earth. In citing it, however, we do necessarily
not endorse it.

You may also like