NO, NO2, NOy Instrumentation Nitrogen oxides were measured by a chemiluminescence instrument built according to established protocols [e.g., Ridley et al. 1988; Carroll et al. 1985]. The NOx and associated PAN components are outlined below. Air inlets for NO, NO2, and NOy, along with the NO2 permeation device, NOy converter and various Teflon valves (Delta; Gulf Technical Service) were housed in a sampling box suspended on a handrail on the Flying Bridge of the atmospheric sampling van. About 5 m of "-OD Teflon sampling lines ran from the box to the van. This sampling box was plumbed to allow NO and NO2 calibration gases to be inserted into the sampling line at a minimum distance downstream of the inlet filters. NO2 calibrate gas was also directed into the NOy sampling line to provide a check on the conversion efficiency throughout the cruise. To avoid contamination by seasalt aerosols, the NO/NO2 sampling line was capped by a 37-mm diameter 1- m pore size Teflon filter (Gelman). Conversion of NO2 into NO for chemiluminescence detection was accomplished within a 0.8-L photolysis chamber illuminated by a 300 watt Xenon lamp (ILC Technology). The up-beam end window was a 3"-diameter Pyrex disk (Esco Products); the down-beam end of the chamber was made from a 3"-diameter parabolic mirror (Edmund Scientific) coated for UV reflection (Al-MgF2 film, Evaporated Metal Films, Ithaca NY). The chamber was designed to have uninterrupted air flow at all times; for all measurements other than the NO2 ambient read and NO2 calibrate, a motor-driven shutter system was employed to block the light path. Thus, for the NO2 blank measurement, the door would be closed. For NOy, a molybdenum converter was used for reduction of NOy species to NO (Joseph and Spicer, 1978; Fehsenfeld et al., 1987). The unfiltered air stream was passed over 8 g molybdenum wire (0.05 mm D) packed in 6" x 3/4" stainless steel tubing temperature controlled (Thermologic) to 400 C. This unit was placed near the sample box inlet to minimize loss of nitric acid. The convert system was evaluated in the laboratory prior to the cruise, where quantitative conversion of NO2 was observed. Unfortunately, the NOy converter became contaminated after 7-Dec (JD 341) and could not be rebuilt. Air for all three chemiluminescence analytes (NO, NO2, NOy) was routed from the box to the van through one sample line; this arrangement has been found to reduce variations in the background count rate. Air flow was controlled by a mass flow controller downstream of the photolysis chamber; photolysis took place near atmospheric pressure and a flow of ~1.2 L/min. Chemiluminescent emission was recorded by a 9658R photomultiplier tube (Thorn EMI) held at 1450 volts and housed in a cooled housing (Products For Research) maintained at -40 C. The housing assembly was mated to a cooled (10 C) internally gold-coated stainless steel reaction vessel (design courtesy of B. Ridley) by an interface of containing a uv filter (Schott RG610). Data acquisition was controlled by a computer employing Labtech software and Metrabyte component boards for control of valves, relays, and motors. A timing sequence was established for the sequential determinations of NO, NO2, and NOy, with a cycle time (including data recording and setup restarting) of ~13 minutes (this was modified during the cruise). Photon counting intervals of 10 seconds were recorded for the duration of the cycle. The first 3-4 minutes of each measure and calibrate count segments were ignored to allow the system to equilibrate. Each run was scanned for spikes or other anomalies due to other shipboard instrumentation; questionable 10-second counts were deleted. For each rate, the resulting 10-second count values were averaged and converted to counts per second. Blank (zero) values were determined by reacting the gas stream with excess ozone prior to entering the reaction chamber. For all analyses, blank values of ~200 cps (before JD 306) and ~400 cps (after JD 306) resulted in a lower limit of detection (LLD) of 1.6-2.3 (leg 1) and 3.0 ppt (leg 2); for a 10ppt measurement the random error would be ~50% (2s). Calibrations for NO were performed for NO, NO2 and NOy; average values for each leg were used for that data set. For NO, a commercial calibration gas mixture (4.11 ppmv NO in N2, Air Products) was injected into the inlet stream. For NO2 and NOy, an 80 cc/min N2 gas stream which passed over an NO2 permeation device (VICI Metronics) housed in an insulated chamber held at 30 C (Thermologic) and added to the ambient air stream. The permeation rate of the device was determined from gravimetric analysis of the device over a period of 6 months to be -26.95 ng/min. Average sensitivities (cps/ppt) for NO, NO2, and NOy during the Transit Cruise were 2.30, 0.88, and 1.78; for ACE1 they were 1.7 0.88, and 1.96. Measured sensitivities varied around these means; excursions were generally related to instrument noise, temperature changes inside the equipment, or to gas contamination from the ship. A total of 2636 sets of NO, NO2, and NOy data were obtained. Raw data for NO, NO2, and NOy were processed into concentrations as follows: 1. The remaining read-blank values were divided by the average machine sensitivity to give the ambient concentrations. 2. Data fields were examined for obvious outliers; these were eliminated. Short-time spikes (e.g., 1-3 consecutive high values) were discarded. Longer episodes were retained even though ship contamination was suggested. 3. Night-time NO values were set to zero by resetting the blank values. In cases were nighttime NO was nonzero due to sampling of ship- affected air, a estimated background value was used. 4. Data were averaged in a moving average of 3 (one before and one after). 5. An estimated error was computed for each time according to Jenkins, 1978 (R. Jenkins, X-ray Fluorescence Spectrometry, ACS, 1978, p.124); these values were averaged as in #4.