Several papers (e.g. [l-91 have discussed the advantages of dividing the effluent stream from a gas chromatographic column between two different detectors and various stream splitters have been proposed. A recent paper [lo] evaluated the use of a "zero dead volume" tee (Valco Instrument Co., Houston,Texas, USA) for this purpose, but severe tailing was observed. Our initial results using thisdevice were disappointing,due towhatwediagnosed as cold trapping in the tee, whose thermal mass is appreciable. By mounting the tee in a small aluminum block contained within the GC oven and heated by a variable transformer, this difficulty was solved and excellent chromatographic results were obtained.Initial attempts employed a Varian 3700 with the Varian split injector; because of discrimination problems, a more linear splitter (J & W Scientific, Inc., Rancho Cordova, California, USA) was retrofitted.Thiswasoperatedatatemperature of2500Candasplit ratio of 1 : 60. The tee was mounted in a 6 X 9 X 2.5 cm aluminum block, containing a 250 W cartridge heater; a variable transformer was used to power the heater and the unit was fitted with a thermocouple readout. The block was initially heated to lOOOC and the temperature then slowly increased to 25OoC, paralleling the temperature of the GC oven. The fused silica column (30 m X 0.25mm,witha0.25~mfilmofDB-5),wasconnectedtothecentral port of the tee, using the standard polyimide ferrule. The relative amounts of the eluting stream directed to the two detectors was controlled by using different lengths (cut from the column) between the tee and the two detectors; 34.5cm was used on the FID side, 9.5 cm on the ECD side; this would indicate a division of ca. 3.6 to 1 in favor of the FID. Great care was taken in cutting the column and tubing ends to ensure square cuts and the absence of polyimide and fused silica debris in the flow path. Table 1 shows the test compounds employed; because of solubility differences, two test solutions were required. One microliter of each solution was drawn into the syringe and the two samples were co-injected. Polynuclear aromatic hydrocarbons were dissolved in dichloromethane; all other solutes were dissolved in hexane. The concentrations of pesticides were all 1 ng/pL; all other solutes were 50 nglpL.