represented by very few samples. Furthermore, if two 
samples of the same species are analyzed during a period 
of deviant values, their reported element contents tend 
to lend credence to each other and, therefore, cannot 
easily be removed from the data collection, although it is 
recognized that their inclusion leads to less accurate 
mean values. Attempts were made, through data check- 
ing, to remove all such deviant values. 
After reviewing the values received for the control 
samples as well as the results of the outlier tests, 
generalizations can be made on the precision of the data. 
Analytical results on controls for Ag and Mn were very 
consistent and very few of the values had to be rejected. 
Controls for Cd were at first reported as zero, but shortly 
after the contractor changed to a dry ash methodology for 
this element, the analytical values for controls became 
very consistent. Very few Cd values were eliminated. 
Likewise, there were few Hg values discarded; however, 
there was slightly more variation among the control 
values for Hg. Zinc showed fairly consistent controls but 
had occasional abnormally high values overall, with 
about 1% being rejected, while Ni, Cu, and Se showed 
more variation in the controls than did Zn but had fewer 
abnormal values overall. During one short period, Ni and 
Zn both had a series of abnormal values. Lead and 
chromium both showed an abundance of zero control 
values for samples analyzed following wet digestion and 
erratic standard values for samples that had been dry 
ashed. There were several periods when almost all of the 
values for both elements were too high to be included in 
the mean, based on the outlier test. Approximately 2% of 
the Pb values and 3% of the Cr values were discarded. 
The Sn controls showed variability similar to Pb and Cr 
but with even more periods of high values. About 4% of 
the Sn values were not used, including almost all of one 
series of 150 samples. Although only about 1% of the 
reported As values was discarded, As showed more varia- 
tion than any other element, with standard deviations 
approaching the mean value itself. Moreover, As control 
values became higher shortly after a change from dry ash 
to wet ash methodology. Results received for Mo, V, and 
Sb differed greatly within themselves over time. Each 
had periods when almost all results were reported as 
zero. When not zero, Sb results showed some degree of 
reproducibility while Mo and V values were more er- 
ratic. 
Explanation of Tables 
A list of all species analyzed in the Resource Survey is 
presented in alphabetical order, by common name, in 
Table 1. The species numbers are those assigned by the 
College Park Laboratory for use in the Microconstituent 
Data Bank and are the same as those used in Tables 2, 4, 
and 5. The first three digits refer to a fish family; the last 
three digits refer to a species within that family. The 
scientific name for each species is also given. 
Areas in which the fish were caught, types of tissue 
analyzed, and the total number of samples of each 
species in the survey are presented in species number 
order in Table 2. Some species were captured in more 
than one area and more than one tissue was sometimes 
analyzed for one species. Specific tissues taken from a 
single species were not necessarily analyzed for fish from 
all listed areas, e.g., black sea bass livers were analyzed 
only for fish collected from the South Atlantic, although 
muscle was analyzed for both North and South Atlantic 
fish. 
Table 3 is in 15 parts. Each part presents a sum- 
marized view of the levels of one of the elements 
measured in the Resource Survey, e.g., the distribution 
of mean mercury level is shown in Table 3.8. The ranges 
of mean analytical values are broken out separately for 
each tissue examined. The number of species, the per- 
cent of the U.S. commercial and sportfish catch in- 
tended for human consumption represented by those 
species, based on 1970 figures (Deuel 1973; Wheeland 
1973), are listed next to the corresponding range. 
Because Table 3 does not identify species, data on one 
tissue may not be compared with data on other tissues; 
Table 4 should be consulted to determine which species 
fall in a particular range and how many samples of those 
species were analyzed. 
The 15 parts of Table 4 are alphabetical lists of 
Resource Survey species, showing the distribution of 
mean elemental levels in the tissues analyzed for each 
species. The columns on the right correspond to the 
ranges of mean analytical values listed in Table 3; the 
range delineating the mean analytical value for a par- 
ticular tissue of each species is marked accordingly. The 
two columns of numbers give the total number of 
samples analyzed and the number of sample values used 
in calculating the mean; “not detected”’ or “‘zero”’ values 
were not used in calculating the mean. 
Table 5 is an MF17 computer printout for the entire 
Resource Survey from the Microconstituent Data Bank. 
Each species is listed separately in species number order. 
The species numbers are the same as those listed in 
Table 1 and are found to the right of the scientific name. 
Within each species, each tissue is listed separately. 
Sites are one degree squares from a specific area (one by 
two in Alaska) designating the location of catch accord- 
ing to the map shown in Appendix Figure 1. The mean 
analytical value is calculated for each unique species-tis- 
sue-site combination. If known, length and weight data 
are tabulated along with the number of samples for 
which the fish were measured. The mean, standard 
deviation and range of these lengths and weights are 
given in meters and kilograms, respectively. This infor- 
mation is followed by the analytical data for each ele- 
ment. Included are the total number of samples from the 
site, the number of samples in which none of the element 
was detected (“‘N Det.’’), together with the mean, stan- 
dard deviation and range of all detected analytical 
values. 
It should be noted that some of the species are 
represented by very few samples and their mean elemen- 
tal values may not be as reliable as those found for 
species more extensively sampled. Where there is a high 
proportion of “not detected” or ‘‘zero” values, their ex- 
