Howard, and White 1971). Very briefly, these 

 modifications were as follows: 



1. Initial digestion of product-30 g of KOH were 

 employed in the digestion, and the methanolic 

 KOH solution was re fluxed for 2-5 h. 



2. Distillation step— 8 g of Ba(OH)o was utilized. 

 The distillation was carried out with the aid of a 

 magnetic stirrer. 



Briefly, the procedure involved digestion of the 

 sample in methanolic KOH, liquid-liquid extrac- 

 tion of an aliquot of the digest with methylene 

 chloride, distillation of the nitrosamines from 

 alkaline solution with further cleanup by solvent 

 partitioning and column chromatography on silica 

 gel and Celite 545 columns followed by GLC 

 (gas-liquid chromatography) analysis. 



A victoreen Model 4000 GLC Chromatograph 

 equipped with a Coulson Electrolytic Conductivity 

 Detector and an Autolab System IV Computing 

 Integrator was employed in the analysis of salmon 

 roe extracts. A 9 foot {2.74 m) x 4 mm inside 

 diameter glass column coated with 10% Carbowax 

 1540 + 3% KOH on 80/100 mesh gas chrom Q was 

 used. The following parameters were maintained 

 throughout all analyses. 



Temperature of injector block - 190°C 

 Carrier gas (helium) flow rate - 70 ml/min 

 GC (gas chromatograph) oven' temperature - 

 ambient for 540 s; GC oven door was closed 

 and brought to 80°C (held at 80°C for 180 s); 

 80°-180°C at a program rate of 5°C/min. 



Conditions of Coulson Electrolytic Conductivity 

 Detector operated in reductive mode were: 



Hydrogen flow rate - 83 ml/min 

 Venting helium flow - 70 ml/min 

 Furnace temperature - 820°C 

 Venting block temperature - 190°C 

 Conductivity bridge - 30 V 

 Attenuation - 1. 



Moisture, nitrite, and chloride determinations 

 were made according to the official methods of 

 analysis of the Association of OflScial Analytical 

 Chemists. 



ducted. A mixture of six A/'-nitrosamines was used. 

 The A^-nitroso compounds were NDMA, dieth- 

 ylamine (NDEA), dipropylamine (NDPA), dibu- 

 tylamine (NDBA), piperidine (NPi), and pyr- 

 rolidine (NPy). Prior to recovery runs, however, 

 the salmon roe samples were examined for A'-ni- 

 trosamines. Several of the cleaner samples were 

 fortified at the 10-ppb (parts per billion) level. In 

 instances where a nitrosamine was found under 

 study, appropriate adjustments were made in the 

 recovery values. Recovery of the A''-nitrosamines 

 at the 10-ppb level ranged from 67 to 88%. 



Representative chromatograms obtained from a 

 fortified pink salmon roe extract together with 

 those obtained from the corresponding unfortified 

 samples are shown in Figure 1. This figure shows 

 the recovery of six nitrosamines after the silica gel 

 cleanup step. Usually, the interferences occurring 

 at a retention time of NPy were removed by 

 further cleanup on the acid-Celite column. During 

 the course of this investigation, blank runs (with- 

 out a salmon roe sample) were made, and the 

 minute GLC peak (3-15 mm) with the same reten- 

 tion time of NDMA observed with all roe samples 

 was not apparent in the blank. As shown in Table 

 1; if the peaks are calculated as NDMA, the levels 

 range from to 3 ppb. Residual nitrite and chloride 

 concentration are also shown. 



A total of 24 salmon roe samples were analyzed 

 in duplicate. All samples contained less than 5 ppb 

 of NDMA. The demonstrated sensitivity of the 

 method was shown to be 10 ppb. A peak with a 

 retention time of NDEA was found (< 1 ppb). No 

 attempt was made to confirm the identity of 

 NDMA or NDEA in any of the samples since all 

 were too low for mass spectrometric confirmation. 

 Some samples were carefully concentrated down 



Results and Discussion 

 During the survey, recovery studies were con- 



FiGURE l.-Gas chromatograms of spiked and unspiked extracts 

 of pink salmon. 



685 



