variation in temperature in different analyses. I n some we get conditions favorable for a loss of alkali, in some for a loss of A1203 and sometimes for both. We have, with great care in heating, obtained only slightly low results where both A1203 and K a 2 0 were very high as in cryolite. The loss therefore probably can be prevented. Boric oxide tends t o alleviate t h e difficulty caused by alumina and i t does not in any way interfere with later determinations. Borosilicates with up t o j

more »

... cent A1203 give fairly good results. One showed 4.82 per cent A1203 and 0.1010 g. alkali chlorides b y regular methods and 4.93 per cent A1203 and 0 . 1 0 2 7 g. alkali chlorides by t h e oxalate method. The results on glass G mentioned in the table are not very good, however. With glasses not containing Bz03 the addition of boric acid before evaporation with H F and H2C20.r decreased t h e fluorine content of the soluble oxalates and also gave higher results for A1203 and alkali without showing any Bz03 present with t h e alkali. From t h e results so far we do not feel confidence in t h e method for such glasses as show undecomposed silico-fluorides, for while accurate determinations m a y be obtained with special precautions t h e method has no advantages under such conditions. For glasses which do not contain much A1203 the method seems well adapted, especially for t h e analysis of a series of glasses of very nearly t h e same composition. We have also found i t convenient for the rapid determination of As203 or Sb203 in glass. D E T E R M I N A T I O N O F B O R I C ACID I n t h e analysis of borosilicate glasses we have found t h a t for the determination of boric oxide t h e method described b y Wherry' is very useful: "Fuse the sample with about 3 g. of Na2C03 for 15 minutes. Take up with 2 0 to 30 cc. of dilute HCl adding a few drops of " 0 3 to oxidize ferrous iron. Place in a 250 cc. round-bottomed flask, heat nearly to boiling, and add dry precipitated CaC03 in moderate excess. Connect with a return condenser and boil vigorously for about I O minutes. Filter out the precipitate through a small Biichner funnel, washing several times with hot water, but keeping the total volume of liquid below IOO cc. Return the filtrate to the flask, add a pinch of CaC03 and again heat to boiling; then connect with a filter pump, through a splash trap, and continue suction until the boiling has nearly ceased. Cool to the ordinary temperature, filter if the precipitate has a red color from iron, add four or five drops of phenolphthalein and run in slowly N / I O NaOH solution until the liquid is strongly pink in color. Introduce about I g. of mannite and shake, whereupon the pink color will disappear: Add NaOH to end reaction, then another gram of mannite and if necessary more alkali until a permanent pink color is obtained." The method was first tried b y fusing 0 . j g. sand and 0 . 2 g. boric acid with 3 g. soda. The melted mass was taken u p with water and 7 cc. conc. HC1 added after transfer t o a 2 j o cc. flask. The method was then followed as described above except titration was made with Ba(OH)2. The B203 content came from 2 to I O per cent low, due t o retention of B203 b y the precipitate' from C a C 0 3 . By using suction in filtering and washing this precipitate practically I O O per cent of t h e Bz03 introduced was recovered. For 0 . 2 j g. B203 = 12.40 B a ( O H ) 2 we used r s . 3 j and 12.40 cc. 1 J . A m Chcm. SOC., SO (1908), 1687. As iron and alumina are rarely present in large amounts in glass and as the time of the soda fusion with glass can be cut t o a few minutes the method seemed quite readily adaptable. The first glass tried was a soda alumina borosilicate which, from analysis, showed a possible B203 content of 10.40 per cent by difference. Three determinations by Wherry's method showed 10.72, 10.j7, 10.67 per cent B203 or an average of 10.64 per cent. A soda borosilicate containing a small amount of BaO and S b~0 3 showed 2 j. 5 7 per cent B203 by Wherry's method and 25.7j per cent by difference. Other glasses of same type showed: Per cent B203 by titration 2 5 . 1 0 12.71 17.57 14.90 Per cent BzOs by difference 2 5 . 1 2 12.15 17.51 15.01 With a zinc borosilicate very unsatisfactory results were obtained, t h e B 2 0 3 content b y titration running from 4 t o g per cent when only 2 per cent mas present. Several mixtures of ZnO and sand with a known amount of B203 were run through by Wherry's method and results for B203 were always high even after very long boiling. Solutions of ZnCI2 and borax were boiled with CaC03 for varying lengths of time and using slight excess and large excess of CaC03 and zinc was always found in t h e filtrate from the C a C 0 3 precipitate. CaO was substituted for CaC03 b u t with no better success. iXa2C03 completely removed the zinc from solution and b y a double precipitation, using first N a~C 0 3 and then C a C 0 3 , we obtained 0.0360, 0.0361 and 0.036 j g. BzO3 when t h e theoretical amount was 0.0365 g. P b O was found t o cause the same trouble as ZnO and the same modification of the method gave satisfactory results. We have been able t o use t h e modified method on a large number of glasses with satisfactory results. The modified method then is as follows: Fuse 0.5 g. of glass with 3 g. N a 2 C 0 3 for one or two minutes after mass is liquid. Take u p with 2 0 t o 30 cc. of hot water and when t h e melt is entirely decomposed filter out a n y insoluble oxides. After washing, transfer filtrate and washings t o a 250 cc. round-bottomed flask, a d d about 7 cc. concentrated HC1, heat nearly t o boiling and add dry precipitated CaC03 in moderate excess. From here on t h e method is as given b y Wherry except t h a t we used Ba(OH), instead of NaOH. It is also advisable t o use suction for filtering t h e C a C 0 3 precipitate.