My experiments with volcanoes

“The Constitution is an experiment, as all life is an experiment.”

At the end of the century my experiments with the sclerometer, and with the class in experimental geology, steered me for years into laboratory experiment. Europe was headed toward geophysics and geochemistry, meaning chiefly mathematical and statistical analysis. My vision was nonmathematical, though I used pressures, temperatures, clocks, and yardsticks to measure erosion, sediment, warping of strata, and melting.

This took me away from petrography, for the polarizing microscope was dealing with infinite series of minerals and molecules. I could see nothing but infinite penetration into the smaller and smaller. Clarence King and Frank Perret had been on the way to infinite journeying outward to the bigger and bigger.

The guiding formula was “erosion, sedimentation, deformation, and eruption.” Measure these on the globe, imitate them with mud pies in the laboratory. Compare the global examples with the mud pies. Try to get the mud pies to illuminate the gigantic stream systems, flood plains, sea bottoms, folded mountains, intrusions, and lavas of the earth. Then try to measure in the field those processes with observatories. So, to me, came the transition from collections to experiments.

The machinery of nature, whether with sand heaps or sand grains or coral pebbles, is the same. It is impelled by currents flowing over loose materials which make eddies in the lee of lumps. The eddies are either billows or cyclones. At the middle they are billows; at the ends they are cyclones. The billow eddies obstruct the heaping. The cyclone eddies lengthen the heaps right and left of the current direction. George Darwin studied the eddies by means of a drop of thick ink in a glass tank on top of a ripple ridgelet. The ink migrated to form underwater billows and cyclones, or vortexes. He used a dropper to place the ink globule, and then watched the vortexes form as he oscillated the tank.

Low parts travel fastest, namely the points. High parts build on the upstream side, and travel slowest, and the stuff tumbles over the crest line and is corniced by the eddy. Snow does it, pebbles under sea do it, and marine life adapts itself to it, wherever the food supply is best.

In the study of ripplemarks, Harry Gummeré, a graduate in astronomy, was my collaborator. Ripplemarks are made by back-and-forth eddies on the bottom, while big waves oscillate the water. We moved the bottom instead of making waves in the water. A glass plate sprinkled with sand under water in a tank was oscillated back and forth horizontally. It was clamped under a carriage which oscillated on wire tracks stretched across the tank. A string pulled the carriage against an elastic on the other side. A wooden wheel and crank, set upright edgeways, had holes and pegs to pull on the string, and the crank turns were timed with a metronome. The holes in the flat wheel were a centimeter apart, so that a revolution of the wheel pulled the string for every two centimeters of travel of the carriage. Thus the sand-covered plate was jerked back and forth under water two centimeters, four centimeters, six centimeters, and so forth; once a second, or two seconds, or three seconds, and so forth, by beats of the metronome.

The result was beautiful ripplemarks on a glass which could be lifted out of the water, dried, and placed over blueprint paper to preserve the record. The sizes of ridge to ridge ripples were from a fraction of an inch to two or more inches. The little ones diminished to zero when the jerking was small, the big ones washed out when the jerking was too big.

The blueprints showed that both length and speed of strokes (amplitude and acceleration of motion) made the ripples increase in size, and somewhere between the largest and smallest sand ripples was the optimum perfection of ripple form. The blueprints look like mackerel skies. And mackerel-sky clouds are billows of condensation between an upper cold stream of air and a lower moist one. In between are the same back-and-forth billows of vortex as in our sand.

At a geological conference at Harvard I showed blueprints made directly from glass plates covered with artificial ripplemarks. At the same time I exhibited rock slabs of fossil ripplemarks and photographs of others shaped like horseshoes. These were variants of the rippledrift process seen on sandy beds of running streams. I also showed photographs of swash marks running along the upper steep slopes of beaches. And of the wind-formed rippledrift of dry sand dunes. From the deserts of Peru come photographs of medaños, or crescent dunes, hills of sand tapering to curved points at both ends. The points are downwind, the high horseshoe toe of the hill is upwind, and like a coral atoll the edifice is current-formed.

Ripplemarks can form in hundreds of fathoms of ocean water if the storm waves on the surface of the sea are big enough. A particle of water on the crest of a wave is lifted up and down in a long vertical ellipse. A particle deep down under the wave is lifted fore and aft in a long horizontal ellipse. Under a three-hundred-foot length of wave in the English Channel in deep water the bottom particles of water are shoving sand back and forth, and making packed ripplemarks.

A big sand grain becomes a lump for small sand grains to bump against. They make a heap which piles up and lengthens out. The heaps merge and we get a tightly packed and ridged sandy bottom. Each ridge has an eddy first on one side, then on the other, as the water particles reverse in direction. Oscillation builds first flocculence, then alignment, then even spacing. The opposite sides of a ridge have equal slopes.

Rippledrift is made by a current in one direction. It is usually not so regular or in such straight ridges as ripplemarks. If a stream of water is jetted over sand round and round in a ringshaped tank, ridges will migrate along the bottom, but they are smeary. The regular ripples in dry sand on dunes have flatter slopes upwind, steep scarps downwind. They are regular, probably because wind blowing is intermittent and back currents occur. So they become more like ripplemarks.

On the bottoms of water streams, the horseshoe rippledrift requires a nice adjustment of lumps and side points migrating downstream. All rippling requires a sand of mixed sizes of grains. If they were all alike they would not ripple, for the larger grains have to obstruct the smaller ones in order to produce the ripple pattern. Rippledrifting as a whole is a building mechanism. Mixed with wave currents which move beaches along, including beach pebbles, it can be compounded into building oceanic islands. The crescent dunes of the desert are dependent on the prevailing winds being loaded with a sand supply at a windward erosion source.

Oceanic currents depend on the winds, like the trades in the tropics, and an obstructing bank or shoal adds surf action to the streaming. If corallines and Tridacna clams and crabs add organic cements, a horseshoe hill is built on the sea bottom. Big eddies will do the same kind of work as little eddies. This phenomenon extends all the way from the galaxies of stars with their beautiful spirals, to the spiral eddies in molten lava rushing down a pit crater, or to the streaming of protoplasm in a plant.

De Candolle, the great botanist, studied rippledrift in order to try to solve the most abstruse problem in all biology, the unsolved mystery of cell division. At some critical point a budding cell decides to form a partition and divide in two. Why or how? De Candolle thought that the protoplasm granules circulating around the cell walls might start regular lumps on those walls, and so build rippledrifts and make eddies.

Thus a current and an eddy and mathematics might start many of the doubles, triples, hexagons, and stars of the world of shells and living tissues. And the cells could pile up in symmetry in the submicroscopic world.

The erosion of the earth’s surface reveals symmetries. River maps look like trees with branches and with rivulets as twigs. Other symmetry is in the horizontal plane of the ocean, where headland furnishes pebbles and the sand sweeps into pure curves of beach and bar and cusp. So a delta builds into a lake of leaf shape and annual layers are added as the flood seasons come.

Some of the fingerlike drainage of erosion cuts into plowed lands during a rainy spell. This suggests what might be done with a spray, a mud bank, and a tank, to see how the finger valleys form. This erosion of the runoff of water was imitated in the Harvard laboratory.

A beautiful river pattern on a slope, like the trickle of raindrops on a windshield, was made by tipping up a rectangular glass plate covered with very liquid clay. A portion clung to the glass, and exquisite fernlike streams formed on the upper half of the plate, with a bank of distributaries of V-shape on the lower slope.

This glass plate was used for a surface of stamp mill slimes, of thicker beds, and was eroded with an atomizer and water by means of a barbershop air compressor. The slimes are very fine pounded sands with angular fragments. To get a stream pattern, this is necessary, so as to have fine grit to cut down the rivulets between the coarse grit remnants. This resembles the requirements for ripples.

The spray was kept going for hours. Meanwhile the river pattern at the steep sides of the sloping plate ate into the bank of sediment, robbing the streams of the main slope, because the side streams were oblique cascades. They dug deep, took off the water, and left the main slope streams without their headwater drainage. The pattern of the main slope became the headwater branches of the side streams, the streams which in plan drained over the edge of the uplifted plate right and left. This was somewhat like stream robbery.

For example, the Lewis River at the south end of Yellowstone Park once drained Yellowstone Lake, including the Lamar River, which is now the headwaters of Yellowstone River. The Yellowstone plateau formerly drained south into the Snake River and the Pacific Ocean. The Yellowstone River headwaters suddenly tapped the system, thanks to geyser erosion and acid corrosion, and the Yellowstone Canyon cut down rapidly, reversing to the north the outlet of Yellowstone Lake. Thereafter the lake flowed into the Mississippi and the Gulf of Mexico. At some critical time about the glacial period the continental divide made a leap of thirty miles from the present head of the Canyon to the neighborhood of Lewis Lake, or from one end of Yellowstone Lake to the other. This is stream robbery.

Spray and runoff and rainfall and wash did not alone cut down the Yellowstone Canyon. The essentials were the rotting of rock and the pull of gravitation on the fragments. The Yellowstone rotted away on the north side, but it was hard granite and mountain-built quartzites on the south, toward the Tetons. Hot spring rotting, geyser erosion, acid waters, and sulfur decomposed the north country. The underground water head followed the easiest channels, and the canyon was the result. The canyon line encircles Mount Washburn, the old volcano, and conceivably is over an old crack concentric to the dome.

Water is a transporter, and cracking opens ways to the rotting agents. Only in rivulets and floods does water actually corrade, or grind, the bottoms of streams. In our spray and fern patterns there is analogy to rainfall springs on flat strata, but nine-tenths of the elements of erosion are left out: jointing, weathering, ice, faulting, gravitation, rotting down, quaking, solution, sliding, and last, spring water.

Erosion by sliding continues by wind action in desert mountains, and on volcanic cones under bombardment, and by rocks snapping under chill and sunshine on the moon. Creep of loose stuff is the greatest eroder on earth. Rainfall cloudbursts certainly help, especially where soil is not held together by a mat of roots.

The process of erosion is supposedly slow, as all geological processes are slow, if we neglect the possibility of such submarine landslips or supramarine upheavals as occurred at Yakutat in 1899. But even New England has floods, hurricanes, landslides, forest fires, and cloudbursts which are exclamation points on an otherwise sleepy history. And in the past it has had ice sheets, and subsidences beneath the sea.

In other words, the making of valleys and stream patterns for the map is accented occasionally, and the occasions may come in climatal waves unknown to us. The stream patterns in the Bad Lands, Tennessee, Pennsylvania, the Grand Canyon, and New England make very different maps. The rotting of the rock, limestone caverns, rainfall, faults, and sloping underground strata bearing spring water all influence these maps. What is erosion and what index is written on the land to say the Grand Canyon and tributaries are being carved downward faster than the Mystic River in Boston?

Ralph Stone tackled the Mystic River, and marked ledges and set stakes opposite the flood plain meanders. The idea was that ledges split by winter freezes, and that the meanders of a stream build on one side and cut on the other. Maps were made repeatedly, and the ledge cracks were measured in millimeters. Some movement was found, but a college year was not enough time. If we could combine as a motion picture, photographs from the air taken once a year for many years, doubtless the film would show that the stream meander pattern is migrating toward the sea like a wiggling snake.

Stone next made a model three inches thick in a tank of water, by sedimenting sixty-one very thin layers of marble dust, coal dust, clay, red lead, and sand. He tipped it up as an island and sprayed it in periods which lasted one to ninety-two hours, up to a total of 719 hours. A forking stream and its delta were formed in the lagoon of the tank. The stream cut a canyon with waterfalls, treelike branches, esplanades, and a flood plain. There were three principal hard white multiple strata layers in the model, separated by sand. The white layers made waterfalls and were eaten back to form the canyons.

When the cross section of the delta was sliced with a knife, it showed three white layers foreset at thirty degrees under the tank pool and separated by more sandy strata. The bottommost of these was the sediment of the top thick marble dust layer of the model as first eroded by the spray, and the top frontal layer of the leaf-shaped delta was the product of the erosion of the canyon bottom on the lowest of the white layers. This must happen in nature where one formation in reverse order is derived by river erosion undermining a stratified older pile of sediments.

We called this the Grand Canyon model, and it showed many features similar to those of South Dakota Bad Lands and the Colorado River drainage. It was strictly rainfall erosion and stratification soakage and seepage. The model surface sloped ten degrees, the high divide at the top had a backslope of forty-five degrees, and everything was sprayed for two months with special hose nozzles, making during part of each day a mistlike rainfall.

The steep backslope did not trench itself at all despite its steepness. This slope, on the contrary, absorbed moisture and carried the rainfall underground down the dip of the strata to add spring water to the main streams. The backslope was a “steep escarpment,” supposed in physical geography to migrate by trenching backward, but the rills never gained volume enough to cut into it. All the water volume acquired its grit for cutting from the large surfaces, which were gradually tilted in the direction of the rivers.

When the complete series of experiments on erosion and sediment was published, it showed that the treelike branching of rivers is dependent on underground water surfaces; that meanders on a flood plain are partly a bubbling-up process of flood-plain soakage; that when side tributaries form by undermining, the upstream branches cut off underground water from the downstream branches; and that when a country is tilted in one direction, there is a tendency to parallel streams, separated by intervals controlled by underground water areas reached by the undermining tracery of headwater springs.

This arborescence in a spray model is a regular and delicate adjustment, where a bunch of tributaries is not mere catchment of rainfall, but is the product of sheet flood in belts of underground water related to the tilt of the country. Arborescence of river drainage on a surface of flat strata, like the coastal plain of the southeastern United States, is a rhythmical pattern of exquisite design capable of reproduction and study in the laboratory. It is a mathematical forking and headward development dependent on volume of water, undermining impermeable strata along permeable ones. And after the “tree” map is formed, the bulb of branches and twigs and underground leaves of spring water holds all the downslope country in its “shadow,” so that no new rivers can form there. This is what makes our great maps of river systems. It is not haphazard. It is a vast ocean of underground water, with mountains of water and valleys of water.

A great lake marks an underground soakage water level. A riverbed marks an underground seepage topography. The sea of water inside a continent is just as much a map of hills and dales of water as the land is a map of the hills and valleys of geography. The water is dynamic, it is flowing. The land surface is dynamic and rain fed; it is creeping soils. Together, groundwater and rivers are melting down the landscape as a living thing. Man dams the water and uses the power of the erosion melting down the land.

When we went to Haystack Basin north of the Yellowstone Park, we found that all of the mountains surrounding it were audibly crumbling. Ultimately, the continent is all one thing: a falling body of rotten rock, ice, water, sand, boulders, and soils, self carved into valleys and mountains, always tumbling. And down below are the fault blocks, prisms of earth shell over the white hot core. And that also is eternally in motion, irrupting, earthquaking, lifting, falling, scraping, heating, cooling in waves through the ages. Man is very tiny, but if he listens he can hear the earth’s heartbeats.

At hot springs the water mantle meets the hot earth shell. So the geyser basins of Yellowstone, California, New Zealand, and Iceland are a hot part of the great erosion system of groundwater. This brings us to the next group of experiments, the making of artificial geysers.

Geysers as eroders show that the under earth is hot and is invaded by rainwater. In exceptional volcanic places the water is boiling hot. The Firehole River of the Yellowstone is carving down basins of solution faster than the regular geysers are building up siliceous sinter. Here is boiling-spring erosion by solution. It may be called the extreme thermal aspect of ordinary spring-water erosion. How does spring water erode? By bubbling up under the beds of rivers. The bubbling out of springs starts rivers, and flood rainfall starts soil gullies; land sculpture is the result.

We introduce geyser experiments here because boiling springs make drama out of ordinary springs, just as active volcanoes make drama out of buried volcanoes. Ordinary springs and buried lavas intruding invisibly are much more important and extensive than geysers and volcanoes. Most people never think of a spring as one of millions bubbling up the beds of brooks and rivers and sea bottoms.

Most people never think of volcanoes erupting—properly speaking, irrupting or inrupting—under Kansas or Brazil. Nobody denies those places are hot underground, but it all seems remote. Yet every spring is thermal if there is heat escaping through the rocks around it.

Geyser basins lower the country around them and leave hills in relief. The proportions of basins and hills depend upon the runoff of rotting and dissolving rock. The shape of a hill standing high, what Davis called a monadnock in New England, depends on its whole history, not on its hardness. Ascutney Mountain stands high as a lump because surrounding slates have rotted down. Mount Monadnock may stand high because the springs under the river pattern of cracks neglected it in the rotting and crunching of a continent.

Dynamic weight eternally falling makes low places. Hardness against weathering makes a mountain high only as a relic or residual. It is a node in the gigantic process of gravitation rotting and the spring squirting of groundwater. The water heats, rises, dissolves, siphons, springs up, and transports dirt. Underneath is a definitely heated earth crust.

Accordance of summit levels of mountains and hills as one looks across country does not have to represent an upraised plane surface. There is more undermining where the spring squirting is most voluminous. When spring squirting is equal, the opposed slopes of a valley adjust themselves. The tree line, the snow line, the rain line, and the wind line are definite levels of erosion. Under it all the rotting rock is falling toward the earth’s center, slowly, creakingly. The everlasting hills are not everlasting, they are everfalling; rocks, boulders, slopes, waters, gravels, sands, and muds. And adjustment to the atmosphere and groundwater surface is irresistible.

The notion of erosion pulling down hills to a flat plane near sea level is fascinating to geometry-minded people, but not to the mechanically minded. A flat plane near sea level in the Mississippi delta is where the river has swung right and left against valley walls, over its own flood plain. A flat plain, secured by ice sheets or planed off by encroaching wave action as land sinks is mechanically probable. In these circumstances we look for river or ice or wave-beach deposits. But an “almost plane” occasioned by the multiple action called erosion down to base level is to me the delightful dream of map students. If a landscape has been planed off, a machine router or planer did it. The great rivers of China have had a long time to bang back and forth against their confining boxes of rock and on top of their own mud.

To return to geyser-spring experiments, I built a simple quart flask surmounted by a four-foot glass tube. At the top the tube rose through a cork in the bottom of a two-foot pan. In the side of the cork of the flask was a second tube with a hose leading up to a reservoir bottle of water. The reservoir bottle could be raised or lowered. If the water in it was level with the pan, there was hydrostatic equilibrium: the pan a pool, the bottle a source, the flask and tube full. When we applied heat to the bottom of the flask, the water boiled, the pan overflowed, and some cold water from the bottle chilled the flask. The pan had become a boiling spring.

Next we lowered the reservoir bottle. The reduced head of water permitted no overflow at the pan, and steam bubbles accumulated in the four-foot upright tube. The boiling point was controlled by four feet of water pressure. If the bubble lift reduced this to three feet, there was a lower boiling point, the pressure was reduced by overflow above, and the whole flaskful boiled. The geyser tube became a regular geyser at intervals of a minute and a half, with eruptions enduring twenty seconds.

This was a miniature of Old Faithful in the Yellowstone. Old Faithful is bigger, its intervals average sixty-five minutes, and they range from thirty-one to eighty-one minutes. It jets up 150 feet for a period of four minutes. It throws out 3,000 barrels of water at each eruption. Our little machine threw up about a pint to a height of four feet.

We hear much about soaping geysers as an artificial stimulus. The apparatus in our laboratory showed the effect of soap right away. When some soap was put in the pan, the intervals of a minute and a half shortened to one minute. Soapsuds accumulated in the tube and depressed the water to the neck of the flask. The multiple bubbles, film against film, made the water system viscous. The myriads of tiny steam bubbles formed so fast that they shortened the lifting time for the column. If the height of the reservoir bottle was so adjusted that the geyser didn’t quite know whether it was a geyser or a boiling spring, the soap made the decision, and the thing went off with a bang.

This simple group of experiments makes springs very real. The Yellowstone explosive springs differ from other springs in having superheated steam from live lavas to heat them. The rock is cracked and the water is doing a job of solution and deposition. It deposits stout tough silica around some openings and builds them up against the head of groundwater, and they become geysers. It deposits lime dissolved off underlying limestone at Mammoth, and this makes sculptured terraces but not explosive springs because the temperatures are not so hot. In both lime and silica regions, blue-green algae, which love hot water, decoratively sculpture the pools.

Like a magician I exhibited the artificial geysers before New York and Boston science academies, and gave the summaries of the results of our geyser experiments, as follows: (1) Boiling springs are like other springs, controlled by the head or pressure of underground water in the hills. (2) Upstreaming of heated water and building up of silica (convection is the scientific jargon) may push the vent of a boiling spring even higher than its source (reversed head). (3) In this delicate condition, even rainfall or sinter building up or outburst at a lower level or clogging of a pipe may change spring to geyser or geyser to spring. There are many more boiling springs than there are geysers, and many more hot springs than there are boiling springs, and the word cold means nothing at all. There may be boiling springs under New York City if you go deep enough. That is why the riot of geyser apparatus is worth thinking about. (4) Irregular geysers overflow continually, regular geysers discharge their waters only during eruptions. Both are methods of feeding rivers, just like any other springs. But there is a lot of volcanic heat underground.

This brings up the question of how much a volcanic eruption is like a geyser. Geologists apply a glib word, phreatic, to Japan’s Bandai Volcano, which blew steam and rocks out of the side of a mountain and dammed a river. Hawaiian volcanoes squirt liquid basalt up a crack with flames and red fume and sulfur gas, and almost no steam at all. The answer seems to be that the Palisades of the Hudson may once have been Hawaiian lava eruptions and, further, that lava is still erupting there if you go down deep enough. New York doesn’t know about it, but it sensed it in 1886, when it felt the Charleston earthquake.

All that Catskill water supply of the great city is in cracks above the level of the deep lava, and extends out under Long Island Sound. If the Hudson fault fissure wiggled a little more than usual, and if the deep lava lowered and pulled down some of the Atlantic water, an eruption like Bandai is not impossible in the Watchung Ridge of New Jersey. This is not likely; but the globe has been through revolutions and cataclysms, and the Watchung explosions might start a new geyser basin. Something like that happened in northwest Wyoming in the Pliocene age, during 11 million years, next preceding the ice ages that began 2 million years ago. And the Yellowstone was the result. We shall see more volcano geysers.

Next, the making of deltas became a hobby in our laboratory, in connection with the old leaf deltas scattered on the New England landscape, partly covered with trees within the grounds of the country villas about Boston.

Delta deposits extend upstream, within the mould of the cavern within ice of the glacial period. Thus the map shows a snake-like ridge of gravel, ending in a maple-leaf flat, with lobate frontal slopes. These slopes were much steeper where the dump of the stream on the delta fell over the beach line at the lagoon or lake level in which the delta was built. This was like the delta shown in Stone’s erosion model.

Stone prospected the idea of torrential deltas in a tank, while E. W. Dorsey and I started a tank imitation of the glacial sand delta. In the glacier, the ice tunnel had been supplied with water by melting through the ice crevasses, just like tunnels seen in Switzerland, floored with sand ground up by the ice. There was thus a torrent pouring along inside an arched tunnel, the mouth of which emerged on a delta in a pool, with water surface either at the tunnel level or above it against the rounded front of the ice mass.

In imitation of a rounded bank of ice with a pool of water in front and with a subglacial meandering cave fed with sands and a torrent, an apparatus was built and supplied by a hose. A sheet of lead was bent in the form of the glacier surface, with an arched opening, and set in our tank. This fitted over a tunnel of sheet iron, soldered so as to meander in plan, and fitted at its upper end with pipe and hose connection. A sheet-iron funnel rose from the upper end of this artificial cavern, wherewith to supply different colored sands to the model subglacial river, represented by the hose jet and iron tunnel. The iron tunnel ended flush with the leaden arch.

The object of the experiments was, first, to set the leaden glacier in a pool of water in the laboratory tank. Next, to jet water through the tunnel, supply sediment in successive colors through the funnel, and let that accumulate on the bottom of the tunnel and in a delta in front of the artificial leaden glacier. The deltas and their sliced cross sections in different experiments represented the noted difference of kinds of sand supply or difference in water level of the pool. In one case the water level was below the ceiling of the tunnel where it emerged from the arch entrance. In another, it was above the cavern mouth, so that water of the cavern stream, debouching from the submerged cavern mouth in the lagoon, spurted up with its mud and made a half crater against the glacier front.

These experiments illuminate the gravel-quarry sections of Massachusetts. In those cuts in eskers (serpent ridges) and sand plains (glacial delta fans) were seen topset beds, or flood wash, or foreset beds at forty-five degrees which are the sublagoon frontal wash, and occasionally backset beds where cavern wash gushed upward.

So our cross sections, cut with a knife in the delta, and the winding cake extending upstream in the cavern showed topset, foreset, and backset strata after draining the tank and lifting out the apparatus. From the embryo delta the flood-plain beds overlap the earlier frontal, or foreset, beds. The frontal beds are always under the lagoon. The flood-plain beds (topset) were made by a meandering river course under the air. Always this plain is built at beach level as a wash fan shaped like a leaf, with the cavern stream bottom as the stem of the leaf.

New England has been covered with mountainous ice, miles high. Subglacial streams and subglacial clear ice caverns are abundantly found at the lower ends of all glaciers in the world. They merely represent the melting snow and ice in pulses of sunshine, snow at the source, ice in the course, crevasses and gravitation making water seep through. This water shapes a channel for itself and erodes a sewer system of scouring along the bottom of the subglacial valley. This grinds and melts the bottom ice into arched caverns; and the sediment builds up on the stream bottoms, eventually carving the roofs of the caverns into high arches or arcades. The subglacial caverns are self constructed drainage pipes.

The glacial stream is really a river flood cutting its valley. The ice river grinds and scrapes, and the water under the ice pipes and drains the melting. The ice carries chisels of broken rock. The enormous weight, in gliding plane layers of ice, flows in accordance with the crystal laws of snowflakes and ice crystals. The moraines, or debris fields, at the sides and on top and underneath the eroding ice jumble yield mud and sand and boulders. The torrent underneath removes the rubbish.

The delta in front follows laws of sedimentation. If there is no lake in front, the delta is a flat wash fan, or valley flood plain. All these things become clear to the student who makes a baby glacier out of tinware, sand, a tank, a hose, and a faucet.

I have spoken of cataclysms, or what early geologists called catastrophe, happening occasionally in the world of erosion and subterranean geysers. Such were the Yakutat crash and the Bandaisan explosion. But each glacier-period field, like an ice mountain over Europe and America, constituted a cataclysm lasting 500,000 years, and this happened four times even in the centuries of early man. The Mediterranean and the Great Lakes are offspring of such cataclysms. But Lyell carried the doctrine of uniformity to extremes; he thought that what man sees is what always happens. I do not believe Lyell ever realized that earth or sun might conceivably explode in a month of our time. Again, this is not likely.

The opposite of uniformitarianism is occasional catastrophic trigger. The process of erosion pulls the trigger for sudden deformation. Slow deformation pulls a trigger for eruption. Eruption triggers internal intrusions. The Frank Landslip; the Charleston, San Francisco, and Napier earthquakes; the Pelée eruption; and the Yakutat upheaval all created terrific surprises for geologists.

The gigantic intrusions through millions of years from the core of the earth, made of white hot star matter, percolating to surface volcano belts up 1,800 miles of permanent, primitive cracks, are mostly balanced by the crustal weight. This is the adjusting globe. But the intrusive mechanism, under tides in the rock and in the oceans, always in motion, pulls the trigger for the big geologic revolutions.

The very deep broken earth blocks shift, volcanism between them heats the surface, floods the surface with gas foam, and lifts areas of surface by heat swellings; and on the surface, what was a glacial period gives place to a volcanic period. The last of these was the Miocene Tertiary, with large-scale volcanic eruptions all over the world.

Comparing Boston with the Black Hills showed underground eruptions in the latter, for which a warping uplift pulled the trigger. These were the rock cisterns or lenses of porphyry injected among the strata. The time of this was Miocene or Eocene Tertiary, probably later than most of the volcanoes of the Yellowstone, farther west.

Boston, on the other hand, was making black basaltic dikes, probably identical with the volcanoes of the Berkshire Hills and New Haven, of the age of the big reptiles, 150 million years before the Black Hills injections. The trigger which pulled off the Boston eruptions was the Appalachian warping. That which fired off the Black Hills was the Laramie revolution that pushed up the Rocky Mountains.

The injection of lava lenses in the Black Hills was a form of deformation of strata which we experimented with in the laboratory. The layers of sandstone, limestone, and old ocean muds covering over the arch of these hills were injected by dikes or fissure fillings from below. How would injections behave?

With Ernest Howe as my associate, I arranged a square tank for sedimenting sand, plaster powder, coal dust, or marble dust in layers under water. Under it was an iron cylinder in which wax could be melted. A screw piston pushed the molten wax up to inclined or upright slots in the middle of the tank box. The water was drained off, and the hot wax was injected up into the strata. The tank sides were taken down, and hardened lenses of wax were sliced vertically with a hot knife to show what had happened to the strata by the process of wax intrusion.

In some of the experiments 300 pounds of shot were piled over a cloth layer on top of the strata to imitate the weight of natural sediments. This was before injection of the hot wax, and the result was a neat dome of deformed layers at the surface, a domical hill over a lens of wax inside. This hill was eroded with a spray of water to show what kind of radial valleys would form. Such radial streams were found in South Dakota, with infacing escarpments, around some of the dome hills made by laccoliths.

From the beginning it appeared that a lens of injection would form, that the strata would arch over a dome of wax. The arched strata stretched on the crest and the breaks gaped upward, while the side bends cracked gaping downward. It was there that the wax could break its way upward and make a volcano. Some nice little experimental volcanoes of wax-built cones and craters formed on top of the model.

As with all folded strata arched downward under weight, the cracks on the bend of a downward arch, or syncline, admit lava from below, whereas the cracks on the upward arch, or anticline, are held tight, closed by the weight of strata above. Thus an intrusive dome will not erupt through its crest, but through its sides.

The results of all these tests showed that rigid beds carried the arching force and that soft beds were most invaded and pushed aside by the wax. The steepness of curvature of arch varied with the load. An inclined pipe formed an irregular lens thickest away from the incline. In a hard bed ruptured on a downbend, concentric fractures around a dome let the lava up to higher strata.

On the crest of a hard bed the fractures are like the spokes of a wheel, but they do not make dikes; they yawn open upward. Liquid wax tended to spread as a thin sheet in soft layers of strata, stiffer wax tended to arch up in a steeper dome. Rapid injection made a higher and smaller dome than slow injection.

Compared with the arching up the whole long mountain oval of the entire Black Hills dome, with granite on the crest, this intrusion of wax only imitates the small domes of lava intrusion or injection, where the injection carries the energy or stress. Indeed, in nature, even the lava lenses are influenced by the buckling that is going on in the strata under stresses of crust warping. For the warping crust of the earth is always pulling the trigger and straining the strata. The lava rising from below seeks out the weak places and assists the buckling, as well as following the most incoherent mud or shale beds.

When it comes to the big oval of the whole Black Hills uplift—swollen up like the Rocky Mountains during millions of years and within which the lava injections were only an item—we are dealing with a push from below or an expansion that swelled up the pre-Cambrian ancient rocks as well as the later granites. Such swellings were doubtless made again and again in Massachusetts. There, also, we find lavas and granites and Red Beds and glacial boulders, older than the Appalachian Mountains, as well as younger. The younger Triassic lavas are definitely erupted between fault blocks.

All that our experiments showed was what melted stuff will do in strata under weight, when the force of melted stuff overcomes that pressure to find a place for itself, although the weight may be more or less lifted by big arching that is taking place on a big scale. The arching is bigger than the hydraulic or gas pressure squirting.

There is another possibility besides buckling. This is faulting, or movement of deep crust blocks the boundaries of which do not appear. The deep crust is a movable mosaic above the core, and this movement renews itself, now here, now there. Dutton shows that we may think of the Rocky Mountains this way all the way out to the Pacific coast. We may have the core fluids sucking down the blocks, the volcanic fluids pushing up the local strata. And the volcanic fluids in cracks are the degenerate gassy top remnants of the core fluids which man has never seen and which are 1,800 miles down.

The boundaries of the crust blocks do not appear because the whole first shell of the globe is buried under lavas and intrusions and crystals and mud, meaning by mud, countless dumpings of lakes and rivers and seas through 3,000 million years. Such is the kind of thinking started by making wax injections.

It will be seen from the experiments that whether we are imitating underground heat with a Bunsen burner to start a geyser, or overground cold with delta apparatus to simulate a glacier, we are dealing with erosion of the earth’s surface. Erosion started with the first attack on lava by the atmosphere or by sea water. Never was the pristine lava anything like the magma inside the globe; it snapped and chilled and oxidized. Whether we call it basalt or obsidian, it degenerated. Moreover, it degenerated in the outer crust when it loosed its gases, heated itself and the rock wall, found groundwater and free air, and started oxidation new to it. Thermal action is just as much concerned with erosion as is rainfall or snow. Therefore, whether injecting wax and swelling strata or imitating geysers and ripplemarks, we were experimenting with volcanoes, for the crust of the earth is fundamentally volcanic. For the purposes of this book these facts demand reiteration.

In what are called geosynclines, or earth sags, the great beds of strata are accumulated. They are the dirt washed from highlands into midland seas. They are strata of sandstone, mudstone, or limestone; thin films in comparison with the fire-made earth crust. It was the wrinkling of basin fills by expansion or end push that built Himalaya and Appalachia. The mountains are etched out of foldings and overthrusts and faults by rotting and water transport. Pressing the strata endways to wrinkle them is called mountain building, much better named strata wrinkling. The thickest of them reached twelve miles vertically, but what is that to the earth’s crust of 1,800 miles? The crust lifts and lowers fault blocks. The little strata basins expand with heat on their bottoms and get pulled and pushed by underground lava intrusions. Also they get squeezed by global contraction between crust blocks, and shoved up and down by the agelong wobbles. The biggest wobble was the downdrop of the great oceans over fault blocks when the crust first cracked and settled over the core. Those oceans have shifted and adjusted in waves of global action ever since. The crust has kept the earth a sphere while lavas erupted and weighted down the blocks. This block wobble extends into the innermost continents. Eruptions up the cracks migrated from the continental seas to the shores of the present oceans. They changed composition as they did so, because they changed from under-air eruptions to under-sea eruptions, fifteen pounds pressure to 600 atmospheres pressure. From erosion eruptions with enormous heat, to deep sea eruptions with enormous chilling and pressure. And the latter are the volcanoes of the present day, mostly concealed except for the islands and sea borders.

Meantime, the crust blocks continue to wobble up and down, and quakes continue to creak under the rock tides of sun and moon pulls. The creaks and wobbles are our big earthquakes, tidal waves, and eruptions. Such big accumulations of eruptions as the Cordillera or the Hawaiian Ridge is a terrific weight in a few million years. Both heaps have been at it since Miocene time, or for about 18 million years, banging down through the crust blocks on top of the core. Whether such balancing of heavy weights on top of the crust blocks is due to change of lava weights or sediment basins, six to twelve miles of rock vertically, the down squeeze and underflow is called by the Greek word isostasy. It means standing level and is a poor word because the earth’s crust never stands still. The blocks are eternally adjusting and creaking over a fluid core, the globe is whirling, the sun and moon are pulling, the volcanoes are erupting, and the solar system is shooting through space. Terra firma is never static. And our little atmospheric lives on top of it never stand still. We are hot, and we ourselves do a great deal of eroding.

This oration is introduction to the next series of Harvard experiments, which dealt with squeezing and wrinkling strata in imitation of the folds and faults of the Appalachian Mountains. Bailey Willis, at the Geological Survey, made a press of wax models of strata. A heavy oak piston was advanced by a screw crank. The models were waxes mixed with plaster for hard strata and waxes mixed with Venice turpentine for soft strata. They were cast to imitate actual successions of hard, thick limestones; less hard sandstones; soft mudstones; or slates. The piston advanced at a measured rate against one end of the model, the other end being a fixed box, the strata lying horizontally. The elongate Appalachian basin had a continent (the piston) to the east; a wide flat fill of limestones or sea bottom to the west (the box); and the deepest trough of pebbles, sands, and muds on the east, toward the rivers of the eroding continent of that ancient time. The heavy limestone tapered from the west into these thinner beds and made a stiff rib in their midst. The final result of their wrinkling was linear folds with axes north and south parallel to the trough, and close set at the east. The folds overturned toward the west, the overturns developing into overthrust fractures westward when the beds ruptured. Also, the folds became bigger, flatter, and wider apart westward under the deeper sea, the famous one being the Cincinnati arch.

The evidence in the middle eastern states is that the trough bottom sank as the heavy shore sediments were dumped by rivers into the sea. The west-central states received a wide flat of limestone. Uplift of the continent shallowed the ocean and pushed it, narrower, over to the great plains. So there were left a deep trough of weak beds, a massive limestone, and an overlap of continental wash across the uplifted later continent of the present time. The problems to be studied in Willis’ models were how folding would affect such a pile, what transmitted the wrinkling force, what started a single fold, and how soft and hard strata behaved under horizontal pressure.

He found that hard, thick layers of limestone transmitted the push farthest. That soft beds piled up on each other near the piston. That these beds showed beautiful overthrust faults inclined away from the piston. And that the start of individual folds was favored by very small initial bends in a transmitting layer. These downbends away from the continent would be made as the trough bottom sank through the ages. The nature of this sinking in upright slices of the bottom rock is probably downfaulting. Each vertical slice would make a step-bend as it sank.

The bottom of Willis’ box did not admit of down motion by underflow, nor did the piston pressure create an opposed horizontal force that might have come from the ocean area. In restraining up motion over the folds that formed, Willis piled bags of shot on top of the model to represent downweighting. The folding in the Appalachians was down at the bottom of the heap where things were hot and compressed, and heat could extend individual strata.

In our pressure chest we extended the Willis conception. We made two pistons at opposite ends of an oaken box, with thick plate glass panes at one side, so as to watch the folding. The two pistons would distribute the end pressure better and admit the possibility that all the pressure did not come from the continent. The bottom under the model was an inner box that could move down, hung on heavy spring balances. These could be screwed up to a pressure upward to compensate the load of shot. Thus the first fold could arch downward as well as upward. This imitated a possible lowered trough bottom. The piston rate of advance was controlled by metronome, one man at each screw.

For examples, models E, F, and G had four white and four black layers, all alike in substance, at fast, medium, and slow rates. The quickest was shortened one inch in five minutes. The slowest was one inch in an hour and three-quarters. The quick-squeeze model flexed smoothly, all folds seemed to flow, and the model held together compactly. The slow-squeeze model shortened the same amount, cracked in many places, was brittle, and did not hold together compactly. This appeared to prove that slow motion will fracture where quicker motion will hold strata intact, under otherwise identical conditions of substance, of folding and shortening, and of vertical confinement.

We verified Willis’ conclusions that stiff and thick beds transmit the pressure farthest and that overthrust tends to form in soft beds, which thicken near a piston. In one model we got overthrusts in opposite directions on opposite sides of the model along a single-fold axis, with a twist in between. While an experiment was in progress, the chest creaked occasionally, the equivalent of an earthquake. One model was cast to represent overlap of strata near shore, like a coastal plain. When squeezed, it made a group of overthrusts away from the piston acting as shore rock.

In burial of strata there is a possibility whereby they wrinkle, and wrinkle most in one direction, which piston pressure does not imitate. That is the heating by burial and expansion or lengthening of controlling layers. In a long basin like the Appalachians, the wrinkling under expansion across the greatest length is easiest, because the axis of stiffness is parallel to the long trough. Transitions off the coastal line from one sediment to the next—sand to mud, mud to lime—will be weaknesses to start bends when expansion pressure takes place under burial along the layers separately heated. These bends develop into wrinkles and the wrinkles, into propagated folds, with the axis parallel to the initial change of weaknesses. Expansion lengthwise on folds, once begun, may make long flat arches pitching in one direction. This heating by burial distributes the folding better and farther than pushing abutments, and makes initial bends. All bottom strata heat and expand in all directions. The direction of easiest yielding to a folding impulse is across the weak transition belts. After that the motion is taken up by linear folds and fractures in one direction.

The models, after continuous or intermittent squeezing, were removed from the chest and sliced with a hot wire for sectioning and photographing. In one, brittle, broken series of folds in a hard layer, the model was taken apart on that layer and the surface photographed. The crest of the folds showed jointing or regular cracks. One set paralleled the fold axes as would be expected; the other set crossed the slopes diagonally and in curves. These last indicated the strains of a twisting nature on a single layer between a downfold and an upfold.

What makes the end thrust, or piston push, in nature? According to the old idea, it was contraction of the inner earth by loss of heat. Willis wrote that the basin sank, isostasy or deep flow was at right angles to the length of the basin, and general contraction took effect by reason of the deep flow. The deep flow was toward the lighter continent, from which the sands were originally lost.

The recent notion that radioactivity heat is in the outer shell denies contraction of the inner earth. Furthermore, I do not believe in a shallow underlayer of lava fifty or less miles down and capable of flowing horizontally under shifting weight. I do believe in a deep underlayer of fluid 1,800 miles down, under a block-faulted crust. This fluid core adjusted itself to the ocean-bottom blocks originally, making the upright slices moving-down controllers of the Appalachian basin. There is no proof that sediment weight did it. It is more likely that igneous, or fire-made, lava, as the thick outer armor plate of the globe erupted in acts of intrusion, lubricated the vertical slices. Intrusions are under every sedimentary mountain range on earth. It is more likely that an agelong up of ridge fault blocks and a down of the basin fault blocks decided where the central continental basin should be, all of it well within the permanent side ridges of North America. For this was a continental mediterranean sea, and the warping of its highland of Philadelphia and its basin of Cincinnati was a mere episode in the 2,000 million year history of Atlantic and Pacific borders of the continent. The sinking of the intracontinental sea, relative to the staying up of the highlands, was a wave in the history of globe and core. Erosion and deposition were results, not causes. They were results of the volcanic history of the ever moving active mosaic of the globe. The permanent North America remained high, relative to Atlantic and Pacific deeps.

The folding of the sediments merges into intrusions of magma in the southern Appalachians. Here arose the granite problem on a tremendous scale, which is repeated in our Ascutney Mountain in Vermont. What it was doing under the bottom of those vast fields of limestone from Ohio to Illinois we have no idea. No more do we know what is doing under the vast fields of lime and red ooze at the present bottoms of the deep oceans. But we do know that fire-made rock squirts up under all sea-laid sediments which anyone has ever studied on islands or continents. This fire-made rock, solidified, has thickness and a bottom. We do not know its thickness nor its bottom. We do know that under it are big cracks 2,000 miles long rupturing it into volcano systems. The conclusion is that the globe is mantled by a layer of igneous matter which has spouted up cracks since more than 3,000 million years ago. How did this matter migrate by new intrusions, to pull, push, heat, and wrinkle through 500 million years the dirt accumulated in shallow Appalachian trenches from Alabama to Indiana? We do not know.

The last of the Harvard experiments that I took part in concerned melting up powders of basaltic minerals and rocks, letting them cool down gradually, and then sectioning them for the polarizing microscope to see how they resembled lavas. V. F. Marsters of the University of Indiana helped me. Based on the European work of Doelter, Fouqué, Michel-Lévy, and others, we used a French furnace with gas flame blast and small crucibles of diatomaceous earth mixed with clay. The specimen powders of crushed natural basalts, or mixtures of pyroxene, feldspar and olivine, were kept glowing for forty to 150 hours, and cooled either rapidly or slowly. The belief in those days was that slow cooling was the main control of coarse crystallization. Quick or slow cooling certainly does produce these effects in lava flows.

From quick cooling, we generally got radial bunches of crystals or spherulites, in a glassy groundmass. From slow cooling, we got diabase structure or coarser crystallization, with some openwork hollow crystals. And there were little grains of magnetite and spinel. Much time was wasted on furnace safety and methods, and on fire-punctured crucibles of platinum, carbon, and graphite.

Nothing had been learned in 1900 about stirring, nor about gas as an ingredient in basalt. It was not until years later, at the Hawaiian Volcano Observatory, that Emerson proved that aa lava was made by stirring a crucible. Aa is crystalline. Emerson got glassy lava by quiet melting. No one has yet subjected lava to hydrogen blasts like those of a Bessemer furnace, nor to other gases. There is a big field here for imitating Mauna Loa and Etna fountains, and for critical petrography of artificial basalts. Modern work has been concerned with physical chemistry of limited mineral systems. So far as I know, no one has mathematically synthesized natural rocks as an object in natural history since the work of Carl Barus for the U. S. Geological Survey in the nineties.