8 BULLETIN- 1325, U. S. DEPARTMENT OF AGEICULTUEE 
For convenience in compiling the data, the quarter-hourly read- 
ings were recorded by means of a typewriter on cards of three differ- 
ent colors, blue being used for the outgoing data, yellow for the 
incoming data, and white for all other records and general notes. 
The differences between the successive readings and other data were 
obtained at a later date by the use of an adding machine. The 
long days, with frequent observations and the dismantling and 
cleaning of some of the gates in the evenings to get the apparatus 
ready for the following day, not only necessitated two observers 
but also rendered it impossible to make compilations during the 
active season of work. 
FACTORS INTRODUCING ERRORS IN THE COUNT 
With any untried experimental mechanism consisting, as this 
did, of many units, each unit in turn possessing parts having defects 
either of construction or of design, it is too much to expect perfect 
accuracy during the first season. Before discussing the magnitude 
of the experimental error produced by these deficiencies, it is desir- 
able so far as possible to know the manner in which this error aris?s. 
1. After observing the passage of about 2,000 bees through the 
early exjDerimental model, it was thought that a tunnel had been 
devised which although allowing only one bee to pass at a time, 
would cause little inconvenience to the insect and Tsould avoid as 
much as possible the scraping off of the load of pollen as the bee 
passes through the device. However, further experience with the 
full set of gates on the hive showed that some of the tunnels would 
occasionally allow two bees to pass at one time. Sometimes the two 
bees would get through and establish only one electrical contact, but 
they might cause two or even more contacts because of the irregular 
manner in which the timnel fell under their weight, the movement 
being impeded by the pressure of their bodies against the stationary 
surfaces of the apparatus. Under normal weather conditions this 
difficulty was confined almost entirely to the outgoing gates, as the 
stimulus urging the bees into the outgoing channels was apparently 
much stronger than that attracting them into the ingoing tunnels. 
By darkening the glass windows and reducing the dimensions of the 
tunnels, this error was reduced to a minimum. Kebuilding all these 
tunnels was a tedious operation, consumed a great deal of time, and 
could be done only gradually, so that the early records show a 
greater error from this cause than the later ones. 
2. Rebounding of the empty tunnel so as to form a second contact 
occurred occasionally if the adjustment of the countei'weight was too 
delicate or if it had been rendered so by debris or pollen collecting 
in the tunnel. 
3. Debris dropped or propolis placed by the bees in the clearance 
between the movable and stationary parts may cause double con- 
tacts by slowing down the speed of the tunnel in its fall; but when 
the quantity of debris became so large that the tunnel ceased to 
function, no error was introduced. 
4. The voltage made necessary by the adoption of the telephone- 
message register unfortunately caused a considerable amount of oxi- 
dation on the surface of the mercury through arcing. In spite of 
frequent cleaning of the surfaces of mercury, this occasionally gave 
