L/C buyer's tip: Don't pay $5,000 for the cabinet A Model 1440 high pressure liquid chromatograph can give you separations and operating convenience identical to practically any $9,000 fancy cabinet competitor. ISCO constant flow 2,000 psi gradient or metering pumps have absolutely pulseless, vibrationfree single stroke piston construction. You don't have to settle for a gas pressure system (with you supplying the gas), or septum in jection. The 6 port sample injection valve permits rapid and reproducible sample in troduction. Stainless steel columns have nonclogging supports and fittings with no dead volume or mixing. The dual beam, high sensi tivity UV photometer has eight absorbance ranges from 0.01A to 2.0A full scale, and in stant w a v e l e n g t h selection of 2 5 4 n m or 280nm. The whole chromatograph as pic tured—pump, column (unpacked), photometer,10cm recorder, and sample injection valve, costs under $4,200. Many addi tional features and options are available, all described in our current catalog. Send for your copy now.
ISCO
ISCO
BOX 5347 LINCOLN, NEBRASKA 68505 PHONE (402) 434-0231 TELEX 48-6453
CIRCLE 1 1 8 O N READER SERVICE CARD
Setit...Forgetit Constant Current or Constant Voltage If long-term stability is important in your electrophoresis work, we suggest you investigate our new power supply. Your setting is maintained within ±0.05% for as long as it takes to complete your separations. Change in line voltage will not affect set voltage or current. Other benefits include full burnout protection and easy-to-read, separate voltmeter and milliameter. Single or multiple chamber use is regulated to 1000 VDC and 200 MA. Write today for cortlptet© d^Wfi-
SEARLE Buchler Instruments Division of Searle Analytic Inc. 1327 Sixteenth Street Fort Lee, New Jersey 07024
Power Supply with Buchler analytical electrophoresis apparatus.
CIRCLE 34 ON READER SERVICE CARD 918 A ·
ANALYTICAL CHEMISTRY, VOL. 45, NO. 1 1 , SEPTEMBER 1973
recombined and monitored by a sin gle detector. One mirror is stationary, whereas the other (Mirror 2) moves. The differences in the distances trav eled by the two beams, cfe-di, deter mines whether constructive or de structive interference occurs at each of the different wavelengths con tained in the incident beam. In the simplest case, Mirror 2 is moved at a constant velocity, and the interfer ence pattern which is generated by the detector can be transformed mathematically to yield intensity vs. wavelength information, which is the desired result of a scanning experi ment. Figure 2B represents a Hadamard transform spectrometer (HTS) (9). The energy from the source is dis persed so that each resolution ele ment can be passed through a set of multislit masks consisting of trans parent and opaque slots. The beam is then recombined ("de-dispersed") so that the total intensity passing the masks for a number of different mask patterns is measured. Measurement of R different intensities passing R different mask patterns provides suf ficient information to resolve R dif ferent resolution elements. The conceptual representations in Figures 1 and 2 can be used to com pare some of the merits of the differ ent approaches. One important pa rameter is the average time each res olution element is monitored by its detector. It is assumed that each de vice produces a spectrum resolved into R resolution elements in a scan period t. In situations involving scanned spectrum devices and nonintegrating array detectors, the average time each resolution element is moni tored is t/R. In situations involving integrating array detectors or either of the multiplex methods, each reso lution element is monitored for the total time t. By assuming Poisson sta tistics, the signal S for a resolution element is proportional to the obser vation time; the random noise signal Ν is proportional to the square root of the observation time (10). It follows that the signal-to-noise ratios, S/N, for the two situations described above are proportional to (t/R)1 2 and t1'2, respectively. Thus, for a given detec tor in which the noise is independent of the signal level, the multiplex and integrating array detector approaches have an advantage in S/N propor tional to the square root of the num ber of resolution elements. This im provement in S/N is obtained at the expense of time resolution. In other words an observation made on a tran sient species at a particular resolution element would be averaged over only t/R time units by any of the nonintegrating devices but would be spread
