When Science Goes Wrong

by Simon Levay

  One potentially troublesome feature of the site was already well known. A geological fault, the San Francisquito Fault, ran southward along the canyon, traversing the prospective site of the dam near the bottom of the sloping western wall of the gorge. Thus, if the dam were built, the fault would lie at the base of the dam’s west abutment (or ‘right’ abutment, according to the convention that the viewer is facing downstream). The fault was believed to be inactive, however, meaning that it was no longer subject to the slow accumulation of stress that might trigger an earthquake.

  The reddish rock to the west of the fault, where the dam’s right abutment was to be located, is a conglomerate – that is, it consists of pebbles and cobbles in a matrix of hardened sand or silt. This particular example is known as the Sespe Formation, or Sespe Red Beds. The rock in this formation, especially in a zone near the fault, is rather crumbly and easily weathered, but its most startling behavior is evident when it becomes wet. When I visited the dam site in 2006, I pulled a loaf-size chunk of rock out of the slope where the right abutment once stood and placed it in the creek. The rock underwent a kind of slow-motion explosion: over a period of about 10 minutes, it gave off bubbles, started cracking, and then fell apart into a heap of stones and mud. In spite of this behaviour, Mulholland conducted tests that convinced him that the foundations of the right abutment would not fail or allow water to percolate through.

  The rock to the east of the fault, which would carry the central section of the dam and its left abutment, is a mica schist. This is a sedimentary rock that has been altered by heat, pressure, and shearing forces so that its constituent grains are flattened, which gives the rock a laminated structure rather like slate. This particular formation is named the Pelona Schist. It is much harder than the rock of the Sespe Formation, but its layered structure gives it the tendency to split and slide along the plane of the layers, like a deck of cards. What was worse, the rock layers were tilted such that they roughly paralleled the slope of the east side of the gorge. At the fault itself, where the Sespe Formation and the Pelona Schist met, there was a layer of claylike ‘fault gouge’, about eight inches wide, that had been generated during innumerable ancient ruptures of the fault.

  Undaunted by the problematic geological features of the site, Mulholland decided to go ahead. As the ‘chief’, his word was as good as law, and any review of his decision within or outside of the water department was perfunctory if it happened at all. As for environmental reviews or state safety inspections, such things did not happen in the 1920s.

  Mulholland constructed the dam from concrete. Most of his prior dam-building experience involved earthen dams, but one year before starting the St. Francis Dam he had begun work on his first concrete dam. This was the dam now known as the Mulholland Dam, which confines the relatively small Hollywood Reservoir on the western fringe of Griffith Park. Apparently Mulholland chose concrete because of a lack of clay – needed to form the water-resistant core of an earthen dam – at the Hollywood and St. Francis sites. The design for the St. Francis Dam was very similar to that for the Mulholland Dam, and both were based on the then-current textbooks of dam design.

  The two dams were gravity-arch dams. This means that they depended primarily on their weight to hold back the water in the reservoir. This weight – a quarter of a million metric tonnes in the case of the St. Francis Dam – thrusts directly downward, whereas the hydrostatic pressure of the reservoir water thrusts more or less horizontally downstream. The resulting combined thrust is aimed obliquely downward. For the dam to be stable against tilting, the combined thrust must be directed into the bedrock within the middle third of the dam’s footprint, not near the downstream edge or ‘toe’ of the dam or, even worse, downstream of the toe. The plans developed by Mulholland’s design engineers met this criterion, of course. An additional safety factor was added by the dam’s arched shape (curved convexly into the reservoir). This had the effect that some portion of the reservoir’s horizontal thrust was carried sideways into the dam’s abutments and thus into the canyon walls.

  Conservative though this design was, Mulholland made it much less so by changes that he ordered after construction got under way. For one thing, he twice raised the height of the dam, from the original 175ft to 185ft and then to 195ft – an 11 per cent increase in height. His aim, of course, was to increase the holding capacity of the reservoir. But Mulholland did not order any compensatory thickening of the dam at its base. In fact, he actually omitted the lowermost portion of the dam’s toe, leaving the dam 20 feet (or 11 per cent) thinner at its base than called for in the design. This latter fact was only discovered decades later by Charles Outland, a local historian who made an extensive study of the disaster and wrote a detailed book about it. Outland spotted the shortened toe by closely examining photographs that had been taken during the dam’s construction. Because the dam was taller, and thinner at its base, than the original design had specified, it was significantly less stable against tilting.

  The dam was completed in May 1926. Filling of the reservoir began during the construction phase but was not complete until March 7, 1928. At that point, the water lapped just a few inches below the spillway. Photographs taken at that time convey an image of graceful strength. The dam’s downstream face, rather than being smooth like most high dams that we’re familiar with today, was stepped. This gave the dam the look of a Roman amphitheatre and emphasised its curvature. Behind the dam, the reservoir spread out into the broad upper San Francisquito Canyon and its side-bays: it resembled a natural and serene lake.

  Day-to-day inspections of the dam were left in the hands of the damkeeper, Tony Harnischfeger, who lived with his girlfriend, Leona Johnson, in a cottage located in the canyon a few hundred yards downstream from the dam. On the morning of March 12, 1928, five days after filling of the dam was complete, Harnischfeger telephoned Mulholland to tell him that muddy water was leaking from the base of the dam’s right abutment. High dams usually leak a certain amount of water, and the St. Francis Dam had already sprung several small leaks during the filling process, but the fact that this leak was muddy was novel and ominous. It suggested that water was not merely passing through the dam but was removing material as it did so.

  Mulholland and his chief assistant, Harvey Van Norman, immediately drove from his downtown office to the dam site, which they reached around 10.30am. They confirmed that muddy water was indeed flowing down the foundations of the right abutment, at a rate of about 15 gallons per second. When they clambered up to the site where the water was emerging from the Sespe Formation, however, they saw that the water was clear. The mud was mixing in as the water ran down the slope and across the embankment of a dirt road. Some water was also leaking from the base of the other, left, abutment, but this water was also clear. Relieved to find that the dam was not being undermined, Mulholland made a mental note to have the leaks repaired at some later date, and he and Van Norman returned to the city.

  The dam broke at two-and-a-half minutes before midnight that evening. There were no surviving eyewitnesses to the collapse, but the exact time could be pinned down because it caused an interruption in the transmission of electric power in lines that ran down the canyon. The interruption was experienced in Los Angeles only as a two-second dimming of lights, but areas closer to the dam were plunged into darkness.

  It is likely that Harnischfeger and Johnson did witness the dam’s collapse before they became its first two victims. Leona Johnson’s fully clothed body was later found trapped amongst the slabs of broken concrete at the base of the dam. Most probably Harnischfeger and Johnson saw or heard some premonitory sign of the collapse, dressed and walked up to the dam to see what was wrong, only to be caught by a cascading mass of water and concrete. Harnischfeger’s body was never found, nor was that of his young son.

  A few survivors did hear or feel the collapse. A motorcyclist named Ace Hopewell was driving up the valley shortly before midnight. The road ran along the canyon’s east wall: at the point t
hat it passed the dam, it was cut into the Pelona Schist only 13 feet above the dam’s left abutment. Hopewell noticed nothing amiss as he passed the dam, but a mile farther up the road he heard a sound as of rocks falling. He stopped for fear of running into a landslide, but the sound seemed to be coming from behind him so he continued on his way.

  EH Thomas was one of the staff of the lower powerhouse, which was situated in the canyon a mile and a half below the dam. His particular job, however, was to attend to the surge tank, high on the east rim of the canyon. The surge tank’s function was to damp out violent changes in hydrostatic pressure caused by the opening and closing of valves; from the surge tank, pipes (or ‘penstocks’) carried the aqueduct water down to the twin generators in the main powerhouse building. Thomas lived with his mother in a cottage not far from the surge tank. At about midnight the two were woken by a strong shock, followed by a continuous vibration. They first assumed that it was an earthquake. Thomas dressed, took a flashlight, and made his way toward the tops of the penstocks. Looking down into the canyon, he saw nothing but rushing water. The powerhouse, if it still existed, was completely overtopped by the flood, which scoured the canyon walls to a height of 120 feet above the creek. Thomas realised that the 28 other powerhouse workers and their families, whose homes were in the floor of the canyon, had most likely been swept away to their deaths.

  Actually, three people did survive. One was Ray Rising. ‘I heard a roaring like a cyclone,’ he said later. ‘The water was so high, we couldn’t get out the front door… in the darkness I became tangled in an oak tree, fought clear, and swam to the surface… I grabbed the roof of another house, jumping off when it floated to the hillside… There was no moon and it was overcast with an eerie fog – very cold.’ Rising met up with a worker’s wife, Lillian Curtis, and her young son, who had had similar narrow escapes, and the three of them waited for the dawn together, but his own wife and three daughters died, along with 64 other members of the powerhouse community. The powerhouse itself, a 65ft-high concrete building, was swept away – only the generators remained in place, half-buried in mud.

  The floodwaters rushed onward. About 15 minutes after the dam broke, the flood reached a cattle ranch near the southern end of the canyon. All the buildings were swept away, along with several of the residents, but a few had been woken by the noise and were able to scramble to higher ground.

  At the southern end of San Francisquito Canyon, the creek joins with those from several other canyons to form the Santa Clara River, which runs 40 miles westward to the Pacific Ocean south of Ventura. Near where the creeks meet lay the small community of Castaic Junction – basically an auto park, with tourist cabins, a gas station, a cafe, and a few other buildings, which served travellers on the highway that connected Los Angeles with California’s Central Valley.

  It took 50 minutes for the floodwaters to reach Castaic Junction, but they had lost little of their ferocity along the way. George McIntyre, the 19-year-old son of the auto park’s owner, was alerted by the rushing, crashing noises coming from the east. He, his father, and their cook went outside to see what was happening. They saw bright flashes from the direction of the Edison station at Saugus, and they concluded that there was something amiss there. Then they watched with amazement as one of the tourist cabins began moving off its foundations. Within moments the father and son were knocked off their feet by the floodwaters. While holding on to each other, the two men were swept away, but not before they caught a final glimpse of George’s younger brother struggling to get through the window of one of the cabins. After a while George and his father, were able to grasp onto a utility pole, but the father was injured, probably by floating debris, and after muttering a brief good-bye to his son he let go his grip. That was the last George saw of him. Eventually, George also let go of the pole and was sucked deep into muddy water. At long last, he found himself back at the surface, half-choked with water and mud. After drifting for some time he collided with the branches of a cottonwood tree, where he was able to hold on until the flood had receded. Besides George McIntyre, only one other person survived from the Castaic Junction community – the cook, who escaped the floodwaters by running to higher ground. All the structures were completely destroyed, and four miles of the north-south highway were inundated.

  By this time, the wider world was beginning to find out about the catastrophe. The initial break on the Edison line had been followed by a cascade of wider electrical failures, as more lines were brought down, switching stations were shorted out and emergency connections were overloaded. Soon the city of Los Angeles, the Antelope Valley and all the coastal cities north to Santa Barbara had lost power. Charles Heath, Edison’s superintendent of transmission, later testified that he had guessed as early as 12.05am that the St. Francis Dam had failed, and that, after telling his subordinates to warn the towns in the Santa Clara Valley, he set out for Saugus in his official car, with siren blaring and red light flashing. He said that he reached Saugus at about 12.45am – just about the same time that the floodwaters were entering the head of the valley.

  Heath knew that 150 Edison workers were sleeping in tents at a construction camp on the bank of the Santa Clara River, eight miles down the valley. He couldn’t reach them, because the road and the bridges were out, so he attempted to telephone the camp from the Saugus substation, which itself was being inundated by the rising water. No one answered the phone and then, after several repeated attempts, the line went dead. The flood, forming a wall of water 40 feet high or so, had struck the camp without any warning. Of the 150 workers sleeping there, 85 died. Few of them had even been able to get out of their tents.

  Further down the valley were the towns of Fillmore and Santa Paula. The destruction of roads, bridges, power lines, and telephone lines had cut them off from any communication to the east, but at 1.30am the night telephone operator at Santa Paula received a call from the coast: the St. Francis Dam, she was told, had broken and floodwaters were descending the Santa Clara Valley. Working by candlelight, the operator began calling local police officers, town officials and the residents of the town whose homes were closest to the riverbed. Two police officers drove through the streets on their motorcycles with sirens blaring. They stopped at every third house or so, warning the sleepy occupants of the oncoming flood and telling them to alert their neighbours and move to higher ground. Soon the entire town was on the move. Meanwhile, squad cars raced up the valley to Fillmore and alerted the population there. As the cars attempted to drive farther east, however, they were stopped by the arriving flood, which filled the entire two-mile width of the valley and was advancing westward at about 12 mph.

  Much of the Santa Clara Valley was used for citrus farming and other tree crops. The groves were obliterated by the oncoming water and mud. In addition, the low-lying portions of the valley towns, especially Santa Paula, were inundated. Houses were carried off, to be smashed up by the roiling water or to be left reasonably intact but several blocks away from their foundations. The bridges across the river were destroyed: the bridge at Santa Paula was demolished just minutes after a police officer ordered a crowd of would-be spectators to get off it.

  As the floodwaters moved westward, they also slowed and spread out. By the time they reached the coast, five hours after the dam broke, they were moving at no more than a jogging pace. The coastal city of Oxnard was evacuated, but that turned out to be unnecessary.

  The total death toll was estimated to be between 400 and 450 persons. This figure may be an underestimate, given that then, as now, many undocumented migrants from south of the border lived in the low-lying areas of the canyons and riverbeds. Though impromptu morgues overflowed with bodies, many of the dead were never found. Some were undoubtedly swept out to sea: bodies washed up on beaches as far south as San Diego. Others were buried under many feet of mud. In addition to the human toll, there was enormous destruction of livestock, buildings, infrastructure, orchards, and farmland.

  Eventually, the city of
Los Angeles settled with most of the victims of the disaster without litigation. Persons who had lost family members typically received $10,000 to $20,000 – or $100,000 to $200,000 in today’s dollars. In one case that did go to court, Ray Rising, the sole worker to survive at the lower powerhouse, was awarded $30,000 for the loss of his wife and three daughters.

  What caused the dam to fail? William Mulholland himself, accompanied as always by Harvey Van Norman, rushed to the dam site in the small hours of the morning after the disaster. As dawn broke, an extraordinary scene revealed itself. A slim central section of the dam still stood in its original place, ranging 200 feet above the empty reservoir like a lone incisor in an otherwise toothless jaw. To its left, another huge section of the dam had slumped downward and across the surviving upright section, shearing off much of its stepped downstream face. This section had broken into three or four giant blocks. But the remainder of the dam, including its left and right abutments, were simply gone – carried as much as a mile down the canyon by the floodwaters. The rocky foundations of the abutments had also been scoured away to a depth of 20ft or more. The left wall of the canyon had fallen away in a giant landslide, carrying a stretch of the roadway with it. Evidently the material in this slide had mixed in with the floodwaters, turning them into a slurry that was dense enough to freight the huge blocks of concrete far down the canyon.

  Surveying the scene, Mulholland immediately suspected sabotage. After all, the aqueduct itself had been the object of dynamite attacks just a few months earlier, so why not the dam, too? Mulholland mentioned or hinted at sabotage as the cause on several occasions, such as at the coroner’s inquest on the victims of the disaster. Some experts presented what purported to be evidence of sabotage, such as the presence of dead fish that were supposedly stunned by an explosion.