354 
PACIFIC SCIENCE, VoL XV, July 1961 
contrast, the lava island in Hilo Bay called Co- 
conut Island was fished up from the sea by the 
demigod Maui (Westervelt, 1916: 28), appar- 
ently not associated in Hawaiian minds with 
the demigods to whom they attributed volcanic 
phenomena. 
DIVERSION SYSTEM TECHNICALLY POSSIBLE 
All who have considered the problem have 
agreed that a solution to containing a lava flow 
does not lie in impounding lava behind a dam; 
the topography is not favorable and the total 
amount of lava that would need to be stored 
cannot be estimated. The solution is sought, 
therefore, in some manner of diverting the 
course of flow. A lava flow following a natural 
channel can be entirely diverted along a chosen 
alternate channel if one fundamental condition 
is met— -the artificial channel must be able to 
carry the lava away from the point of intercep- 
tion as rapidly as it is delivered there by the 
natural flow. 
The average gradient of such a diversion 
channel will, of necessity, be considerably less 
than the average gradient of the natural slope 
across which it is constructed. To offset the un- 
favorable loss of gradient, the built channel 
must offer less obstruction in its floor, such as 
irregularities and vegetation, and provide space 
for a greater cross-section of flow. It is not 
enough to consider that a cross-section of a 
diversion channel is adequate by allowing an 
added area to compensate for the reduced gradi- 
ent on the basis only of gravity flow of a liquid. 
Allowance must be made also for the capacity 
of the lava flow to transmit enough heat to 
maintain its liquidity. This differential term in 
the equation works against a wide flow, even 
though its greater width might sufficiently com- 
pensate reduced depth for water. If these condi- 
tions are met, the channel will direct the move- 
ment of the flow, and the barrier need only 
confine the downslope margin, not act as a dam 
across the flow. 
However, if lava behind the barrier is ponded 
to a considerable depth (50 ft. or so), the pos- 
sibility that it might inject itself through the 
barrier or its foundation cannot be overlooked. 
Such an engineering accident was responsible 
for the early failure of a barrier constructed dur- 
ing the I960 eruption of Kilauea. 
DESIGN OF THE DIVERTING SYSTEM 
The designer of a system of structures to di- 
vert flowing lava must know the probable maxi- 
mum rate of delivery of lava that can be ex- 
pected to enter the system. Here again, geologic 
experience cannot predict the probable require- 
ments, it can only point out the possible maxi- 
mum load. If the outbreak takes place within 10 
or 15 mi., lava may be expected to enter the 
system at a rate of about 25,000,000 cu. yd/hr, 
based on the observations made on the Mauna 
Loa eruption of 1950, the most voluminous 
eruption that has been sufficiently documented 
(Finch and Macdonald, 1953). Should the de- 
signer anticipate the voluminous load from a 
nearby eruption? What are the data upon which 
to make the decision? 
The pre-Hawaiian lava flows that form the 
south shore of Hilo Bay (see Fig. 1) appear to 
have come from vents along the lower part of 
the northeast rift, according to current studies 
of recent air photographs and some reconnais- 
sance field identifications. The topographic ridge 
built by these and similar eruptions is the south 
boundary of the topographic trough that slopes 
into Hilo Bay. Any future eruption along this 
rift line below an altitude of about 3,500 ft. will 
lie on the south side of the ridge, and its lava 
flows thus would be directed away from Hilo 
Bay; an eruption along this zone above about 
3,500' ft. will be more than 15 mi. from Hilo. 
Any source vent closer to Hilo than 15 mi. 
would have to break through the flank of Mauna 
Loa considerably to the north of the zone of old 
cinder cones that mark the lower part of the 
northeast rift. However, Stearns and Macdonald 
(1946: 70) reasoned that the vents in Hilo 
(Halai Hills) lie on a branch of the northeast 
rift, and Macdonald restated the supposition in 
1958 (p. 259). An eruption on any part of this 
supposed branch of the rift zone will be in the 
trough leading to Hilo; such an eruption must 
be expected geologically, even though there are 
no existing vents along this line between Hilo 
(Halai Hills) and a point 22 mi. from Hilo at 
