SECT. 4] LIGHT 415 



Even when known, it is often difficult to obtain phototube-filter combinations 

 that will duplicate a desired bandwidth. Added to these difficulties is the fact 

 that ocean water itself limits bandwidth (Tyler, 1959) and in great depths may 

 entirely control the bandwidth associated with the measurement. Current 

 thought with respect to the selection of bandwidth is divided according to the 

 requirements of individual problems. 



The physical approach to the selection of bandwidth is to adopt a mono- 

 chromatic criterion. Monochromatic measurements are the most generally 

 useful since, once available, they can be applied to a variety of problems. 



A second approach to bandwidth selection has been to match the instru- 

 mental bandwidth with the acceptance spectra or utilization spectra of the 

 organism or image-forming system being studied. The recent discovery of 

 the accessory-pigments effect (Emerson et al., 1957) in the photosynthetic 

 action of certain plankton species has emphasized the importance of this 

 approach. 



A third approach to bandwidth selection has been the use of a broadband 

 sharp cut-off filter which limits the bandwidth of the instrument to that 

 region of the spectrum where the attenuation coefficients of the water are 

 known to be lowest. Near-surface measurements can in this way be made to 

 correspond closely in bandwidth to the deep measurements. 



The state of polarization of the flux in the underwater light field is, of 

 course, related to the state of polarization of the flux from the sky and to the 

 subsequent scattering effects in the water. Recent measurements by Ivanoff 

 and Waterman (1958) of the amount and direction of polarized light under- 

 water indicate that polarization exists at considerable depth and should be 

 given proper consideration in the design of instruments. Of the remaining two 

 parameters, magnitude and direction, a word should be said about magnitude. 

 In light measurements in the sea, instrument readings are related to radiant 

 flux by a factor composed of circuit constants, which are generally designed to 

 remain constant, plus an optical coupling factor which must remain constant 

 for all aspects of the light field. 



For an irradiance collector, for example, the optical coupling is stated by the 

 cosine law : 



Jq = Jq cos d. 



A collector plate with an error that becomes progressively worse as 6 in- 

 creases will not yield readings which are directly proportional to irradiance for 

 all sun angles because the direct rays from the sun, which would be the major 

 determinant in the magnitude of the irradiance, will be weighted differently 

 at different angles. Similarly a spherical collector will not yield readings 

 directly proportional to the spherical irradiance for all light fields if the optical 

 coupling factor is not a constant for every aspect of the sphere. 



Optical equipment should obviously be carefully designed and tested to 

 insure a constant coupling between collector and detector. 



The physical quantities and properties which are most important for the 



