electrodes have been mistreated in some way. 
When the zero has been determined the Y 
is drained, the electrode chambers are con- 
nected directly to the sea, and the recorder 
output should give true sea-voltages without 
correction. 
Excessive wave signal can be eliminated 
by crossing the inputs with a capacitor. One 
of the few advantages of the high impedance of 
the salt bridge circuit is the small amount of 
capacitance required to give a long time 
constant. 
If a salt bridge line is relatively dry 
and clean, it will pick up an external electro- 
static charge when brushed or touched, which 
will show up as a sharp pip on the chart record. 
This is not usually objectionable. 
THE PROOF OF THE PUDDING 
Our first successful measurements using 
the salt bridge technique were made on a 
stationary line strung across the bottom of 
Woods Hole harbor from the home of Allyn Vine 
on Juniper Point to the dock on Mink Point, 
Nonamesset Island; a distance of 875 yards. The 
line itself was about 1000 yards in length, 
weighted with 10 1b. sash weights every 16 ft. 
(When we later took up the line we decided that 
one such weight every 100 feet would have been 
ample, except in shallow water.) We made a 
continuous recording, with several interruptions, 
from July 21 to August 9. 
On several occasions the signal was blocked 
by bubbles, which probably could have been 
prevented if we had used bubble traps. On 
another occasion the pump was inadvertently and 
disastrously connected to an electrode chamber 
instead of to the short line which was to be 
pumped free of bubbles. We had to close down 
for two days while we repaired the damage and 
modified the design to withstand future mis- 
takes of this sort. Twice the chart recorder 
ran out of ink. Otherwise the operation was 
nearly trouble-free. 
We were distressed, at first, by the large 
noise level in our signal. There were large 
fluctuations a millivolt or more in magnitude, 
which often continued for many minutes or even 
hours with no visible physical cause. Sometimes, 
especially in the early hours of the morning, 
the fluctuations were beautifully periodic with 
two or three minute cycles for a half an hour 
at a time. Sometimes there were no fluctu- 
ations at all. 
When we showed our records to Dr. von Arx, 
he immediately identified these disturbances as 
being due to earth currents associated with 
178 
magnetic storms. In fact, his original paper 
on the G.E.K. warns of the possibility of such 
interference "in shoal waters near the land". 
When our noise level was compared with the 
Magnetic noise level reported for each three- 
hour period by the Fredericksburg Magnetic 
Observatory of the U. S. Coast & Geodetic 
Survey, the agreement was overwhelming. 
The record reproduced in Figure II was 
chosen as being the quietest 12-hour tidal 
cycle which we obtained. It also happens to 
be the quietest such period in the magnetic 
records. While it is atypical in this respect, 
we believe it to be a fair reflection of the 
intrinsic precision of the instrument and the 
method. 
The pronounced disturbances occurring be- 
tween 1715 hrs. and 1830 hrs. on July 30 are 
recurrent with a 24-hour period, as can be 
seen by a comparison with the juxtaposed record 
of the corresponding period on the preceding 
day. They are caused by the ferryboats which 
operate out of Woods Hole to Nantucket and 
Martha's Vineyard. Like most large vessels, 
these steamers produce strong electrical fields 
in surrounding water due to bottom corrosion 
or electrical leaks. The sharp pulses corres- 
pond to ship passages across our line. The 
half-hour long prolonged 2-millivolt downward 
shift is due to the steamer Nantucket, as she 
stands at the wharf 250 yards away with her 
stern facing us, waiting for the Islander to 
leave the ferry slip. When the Islander leaves 
(sharp downward peak), the Nantucket is warped 
around into the slip and her signal vanishes. 
The later downward peak is the Nantucket 
crossing our lines going out. 
The R.V. Chain creates a similar electrical 
signal when she is in port, which is why we 
did not run our line from the Institution dock 
in the first place. The electrode zero which 
our instrument permits, is of small value when 
an unknown stationary signal of such origin 
exists in the water. 
To test our zero, we examined our measured 
voltages at the times of slack water in Woods 
Hole channel as predicted in Eldridge's tables. 
During our first week, while the Chain was at 
sea, these averaged to zero within a tenth of 
a millivolt. After the Chain. returned, the 
average was off zero by a steady 2 millivolts. 
In order to get the true zero for the tidally- 
produced voltages, we were obliged to use these 
averages to correct for the non-tidal potential 
drop across the harbor. The true zero, thus 
estimated, has been drawn in as a heavy hori- 
zontal line in Figure II. 
In Figure II the full vertical scale is 
20 millivolts. Positive signals are due to 
water flowing westward through Woods Hole into 
