Movenert of the cell sear rniwetok to a position southeast of Majuro gave strong socthwesterly flow in the local erea on 28 Pebruary en? west south west flow on ] Larch, replacing the light variable winds wich existed on 23 - 25 February. Contimed southward ootlor en4 a&prroach of following cell from west brought west to -orthrest flow on 2 and 3 March. (4) At 50,000 feet, the slow movement of the clockwise ci~m culation from a position northwest of Exiwetok to south of Majuro changed the flow gradually from north northrest to west southwest. Aer 1 Uarch this circulation continued its movenent to the southeast and weekened, vith flow in the local area becoming west northwesterly. (5) At 60,00C feet, no clearly defined trend was presext. {Ns ‘winds were cuite light and erratic at this level, with easterly flow at higher levels. These variable winds persisted through PRAVO plus 3 days. ad. Differences between forecast and cbserved winds: In regacd to the differences between forecast and observed winds, reference is cade toa report of Froiest 4.5, JIf THxa=, by Pa-mer, Lilier, and Stopinski. This report stcted that studies of the observa- i tional errors in upser wind ccservetions teen by GD/1 ezsimem indicated thet for wind speeds above 10 knots, arprexiratelr 1f per cent of the observations varied by sore than 20 degrees, eves when the cbhservations were cade dy several Cifferent Grows =ots, but vith the same aizberne unit. The cifferences betmeer Forecast goemeipiaitonn wis en ee ree or incividual cells drifting ezstrard. — ae