INSTRUMENTS AND TECHNIQUES FOR METEOROLOGICAL MEASUREMENTS 
better than the figures quoted above; a probable error 
of --0.5 mb between contacts can be assumed. 
Temperature. The temperature-sensing element is a 
rod of ceramic material, 0.020 in. in diameter and 1 in. 
er te te IE A TE TT 
E i | 
ee ee 
oo . 
4 
pac =| 
Fie. 3.—Complete radiosonde showing the meteorological 
modulator, the exposed temperature element, the humidity 
shield, and the 1680-me transmitter. 
in length. These thermistors possess very high negative 
resistance coefficients, the resistance varying from ap- 
proximately 20,000 ohms to 2,000,000 ohms for the 
temperature range of from +60C to —90C. In order 
to decrease errors due to solar radiation, the elements 
are coated with a suitable lead carbonate pigment 
which has a reflectivity coefficient of about 0.92. The 
coating increases the diameter to about 0.035 in. The 
experimental results of Brasefield [6] indicate that such 
elements may be directly exposed to solar radiation 
without introducing an error in excess of 0.5C, except 
at altitudes above 80,000 ft where errors from 1.0 to 
1.5C may occur. For these altitudes, however, cor- 
rections can be applied. If such corrections are applied, 
it is estimated that the probable error of the residual 
is approximately --0.2C. 
Another important parameter of a temperature ele- 
ment is its lag constant, defined as the time required 
for the element to respond to approximately 63 per 
1209 
cent of the interval between an initial temperature and 
a final stable value. For the new white temperature 
element this constant is about 3.5 sec at sea level and 
for a ventilation rate of 1000 ft min. At an altitude 
Fria. 4.—The selector switch with pressure capsule. 
of 100,000 ft the lag is about 22 sec proportional to 
p °4, where p is the density. If the temperatures are 
corrected for lag, then the residual lag error intro- 
duces a probable error of about --0.2C in the tempera- 
ture determination. 
To obtain a reasonable estimate of the over-all prob- 
able error in determining temperature, it is necessary 
to consider the following sources of random error: 
solar radiation (0.2C), residual lag (++0.2C), trans- 
mitter error (+0.2C), lock-in error (+0.1C), calibra- 
tion error (--0.2C), recorder error (+40.2C), and finally, 
evaluator error (-0.2C). From the theory of the prop- 
agation of errors, it follows that the over-all probable 
error in temperature measurement is about --0.5C. 
Humidity. Unquestionably the weakest link in the 
present radiosonde is the humidity element. The ele- 
ment employed is a modification of the original Dun- 
more [13] design and consists of a polystyrene strip 
4 in. long, 2 in. wide, and 1¥¢ in. thick, coated with an 
electrolytic film of lithium chloride dissolved in poly- 
vinyl acetate or alcohol and provided with two elec- 
trodes along the two long edges. At room temperature 
the resistance of the strip varies from approximately 
ten million ohms to five thousand ohms for a relative- 
humidity change of from 15 per cent to near saturation. 
In addition, the resistance of the humidity strip varies 
with temperature as well as vapor pressure, and ac- 
