Separation and Fractionation of Sediments from Seawater
81
does the 0.8 Jlm silver filter. Presumably the difference results from the larger number
of small particles retained by the glass fibre filters.
The method of calculation of the blank can also influence the determined sediment
content. If surface sea water is filtered through a pad consisting of two or more filters,
either glass fibre or silver, the bottom filter will often contain a small amount of
sediment above the blank value. Some workers have maintained that this is due to the
adsorption of dissolved organic matter on to the filter and that this value should
therefore be subtracted from the weight of sediment found on the top filter (Banoub
and Williams [696], Menzel [697]). Other workers feel that the material caught in the
second filter is largely composed of smaller particles passing through the first filter.
Depending on the way in which the particulate fraction is defined, the material caught
by the second filter should either be added to that collected on the first filter (Bishop
and Edmond [698]) or ignored (Gordon and Sutcliffe [699], Sharp [700], Wangersky
[701]). It can easily be seen that the choice of blank calculation can cause a considerable difference to the final values given for sediment content, at least in surface waters.
As far as the particulate fraction is defined, not in terms of particle size, but in terms
of material caught on a specific filter, it is recommended that only one filter, rather
than a pad of two or more, be used, since the material caught on subsequent filters is
irrelevant by definition.
When uniform methods of collection and analysis are used, the deeper layers of the
oceans give remarkably consistent results. Replicate samples, taken with a Niskin
rosette sampler rigged to close six 5-1 bottles simultaneously, displayed a standard
deviation of ± 1.3 Jlg of carbon per liter (Wangersky [701]).
Methods which collect a greater proportion of the smaller particles have also been
employed. For example, a layer of fine inorganic particulate matter deposited on a
filter of coarser porosity has been used to separate the particulate from the dissolved
fraction. Thus, Fox et al. [702] used layers of calcium hydroxide and magnesium
hydroxide, while Ostapenya and Kovalevskaya [703] used powdered glass. These
filters suffered from three disadvantages - they were troublesome to construct, the
nominal pore size was irreproducible, and adsorption of truly dissolved material was
possible. The techniques were abandoned with the advent of the first membrane
filters having graduated pore size.
Many sediment collection methods used are biased towards those particles falling
very slowly. If the residence time of a particle in the water column is only a few days,
the probability of being caught in a 5-1 Niskin bottle is small. This has been pointed
out by the work of Bishop et al. [704]. These investigations used an in situ pump and
filtration apparatus to filter very large quantities (5-30 m 3 ) of sea water and caught
many classes of particles never seen in Niskin bottle samples. Their results are not
comparable to those obtained in other filtration methods.
4.1.2
Separation by Centrifugation
A method for removing particles, which is not limited in volume sampled, and which
suffers less from problems of overlapping classification, is continuous-flow centrifugation. Separation into density classes can be achieved by choice of speed of centrifu-
81
does the 0.8 Jlm silver filter. Presumably the difference results from the larger number
of small particles retained by the glass fibre filters.
The method of calculation of the blank can also influence the determined sediment
content. If surface sea water is filtered through a pad consisting of two or more filters,
either glass fibre or silver, the bottom filter will often contain a small amount of
sediment above the blank value. Some workers have maintained that this is due to the
adsorption of dissolved organic matter on to the filter and that this value should
therefore be subtracted from the weight of sediment found on the top filter (Banoub
and Williams [696], Menzel [697]). Other workers feel that the material caught in the
second filter is largely composed of smaller particles passing through the first filter.
Depending on the way in which the particulate fraction is defined, the material caught
by the second filter should either be added to that collected on the first filter (Bishop
and Edmond [698]) or ignored (Gordon and Sutcliffe [699], Sharp [700], Wangersky
[701]). It can easily be seen that the choice of blank calculation can cause a considerable difference to the final values given for sediment content, at least in surface waters.
As far as the particulate fraction is defined, not in terms of particle size, but in terms
of material caught on a specific filter, it is recommended that only one filter, rather
than a pad of two or more, be used, since the material caught on subsequent filters is
irrelevant by definition.
When uniform methods of collection and analysis are used, the deeper layers of the
oceans give remarkably consistent results. Replicate samples, taken with a Niskin
rosette sampler rigged to close six 5-1 bottles simultaneously, displayed a standard
deviation of ± 1.3 Jlg of carbon per liter (Wangersky [701]).
Methods which collect a greater proportion of the smaller particles have also been
employed. For example, a layer of fine inorganic particulate matter deposited on a
filter of coarser porosity has been used to separate the particulate from the dissolved
fraction. Thus, Fox et al. [702] used layers of calcium hydroxide and magnesium
hydroxide, while Ostapenya and Kovalevskaya [703] used powdered glass. These
filters suffered from three disadvantages - they were troublesome to construct, the
nominal pore size was irreproducible, and adsorption of truly dissolved material was
possible. The techniques were abandoned with the advent of the first membrane
filters having graduated pore size.
Many sediment collection methods used are biased towards those particles falling
very slowly. If the residence time of a particle in the water column is only a few days,
the probability of being caught in a 5-1 Niskin bottle is small. This has been pointed
out by the work of Bishop et al. [704]. These investigations used an in situ pump and
filtration apparatus to filter very large quantities (5-30 m 3 ) of sea water and caught
many classes of particles never seen in Niskin bottle samples. Their results are not
comparable to those obtained in other filtration methods.
4.1.2
Separation by Centrifugation
A method for removing particles, which is not limited in volume sampled, and which
suffers less from problems of overlapping classification, is continuous-flow centrifugation. Separation into density classes can be achieved by choice of speed of centrifu-