51 



et Thermometer, inddelt i Delstreger af Værdi 0.°2 med 

 en Længde af 0.68"""., hvis Correctioner bestemtes ved 

 gjentagne Sammenligninger med det herværende meteoro- 

 logiske Instituts Normalthermometer, som til det Brug 

 velvillig' lilcv mig laant at' Institutets Bestyrer, Professor 

 Dr. Mohn. 



17.°5 

 1775 

 17.-5 

 1775 



0° 



0° 



af VU 1.02669. 



af 

 af 



t 1.02691. 

 I 1.02845. 



Af Observationerne 32 til 37 kan forst Udvidelsen 

 af det til Pyknometrene benyttede Glasrør beregnes, og 

 man erholder, naar de af Wiillner 1 beregnede .Værdier for 

 Kviksølvets Udvidelse lægges til Grund. som Udtryk for 

 Glassets midlere Udvidelsescoefficient mellem 0° og 15" 



0. 10267 og mellem U° og 20° 0.0000274. Man kan 



desuden ogsaa benytte Observationerne 3. 4 og 5 i For- 

 bindelse med 24 og 25 til Beregning af Glassets Udvidelse 

 og erholder, naar man anvender de 'af Hallstrøm- og Ro- 

 setti " bestemte Værdier for Våndets Udvidelse. meget vel 

 overensstemmende Tal. som i Middel fastsætte Glassets 

 midlere Udvidelsescoefficient mellem 0° og 17:. i til 0.0000275. 

 Efterat man heraf har bestemt en passende Værdi for Glas- 

 sets Udvidelse, hvorved naturligvis de med Kviksolv ud- 

 førte Bestemmelser fortrinsvis maa komme i Betragtmng, 

 kan man nu skride til Udledelsen af de videre Resultater 

 af Forsøgene. Man maa imidlertid her tåge Hensyn til. 

 at Observationerne 1, 2. 10, 26, 27. 28 og 29 tydeligt 

 bevise, at Pyknometret under Forsøgene har tabt omkring 

 <>.7 Mg r. i Vægt; hvad der rimeligvis skriver sig fra Op- 

 losning af en Del af Glasset paa de ydre Vægge paa 

 Grund af den stadige Omrøring i Badet. Den herved 

 foraarsagede Usikkerhed kan dog betydelig reduceres. naar 

 man beregner Pyknometrets Vægt til enhver Tid under 

 Forudsætning af, at Vægttabet er proportionalt med Ob- 

 servationernes Antal. idet Usikkerheden da knapt nok vil 

 influere paa Ste Decimal. Under denne Forudsætning har 

 jeg senere beregnet følgende Resultater, idet jeg paa en- 

 kelte Steder bar tilladt mig smaa Aproximationer. som 

 imidlertid kun kan virke paa de endelige Værdier med en 

 liden Feil i Gte Decimal. 



Egenvægt ved j^"? af III 1.01739. 

 1 1. 1 . > 



1 Pogg. Ann. 153 — 440. 



2 Disse Værdier ere ogsaa af Ekman benyttede til Bestemmelse 

 af Udvidelsen af det Dilatometer. som denne til sine Forsøg' over Sø- 

 vandets Udvidelse har anvendt. 



3 Ann. Chim. Pliys. [4] — 17 — 372. 



too, at least once for every other determination,. and again 

 filled witli the sea-water the volume of which had to be 

 ■found. These observations veie in qo case considered 

 satisfactory unless the temperature of the water-bath bad 

 hoen kopt comparatively uniform for at least 15 minutes. 

 the greatest variation indicated by the thermometer during 

 tbat interval never having exceeded 0.°1. For reading the 

 temperature of the water-bath, a thermometer graduated 

 in divisons of 0.°2, measuring each 0.68"™, was made use 

 of. the instrument having been previously corrected by 

 frequent comparison with the standard thermometer of the 

 Norwegian Meteorological Institute, which the Director. 

 Professor H. Mohn, bad kindly lent me for tbat purpose. 

 Nos. 32 — 37 are the first of the observations by 

 which the expansion of the glass in the pycnometer may 

 be computed; and. taking the valnes found by Wiillner 1 

 for the expansion of mercury as the basis of calculation, 

 the mean coefficient of expansion of theglass betweenO°and 

 15° will be 0.0000267, and between 0° and 20°, 0.0000274. 

 Moreover, Nos. 3. 4. and 5. in conjunction with Nos. 24 and 

 25, will also serve for determining the expansion of the 

 glass; and. adopting the values computed by Hallstrom- 

 and Rosetti 3 for the expansion of water, very closely 

 agreeing results will be obtained. the figures expressing 

 the average mean coefficieut of expansion of the glass be- 

 tween 0° and 17.°5 being 0.000(J275. After a proportionate 

 value of the expansion of the glass has been found. 

 for computing which preference should be given to the de- 

 terminations performed with mercury. we may proceed 

 to deduce the further results of the experiments. Mean- 

 while, regard must be had to the fact. of which the obser- 

 vations 1, 2. LO, 26. 27. 28. and 29 afford conclusive proof. 

 tbat the pycnometer had lost about 0.7'"'"' in weight during 

 the progress of the experiments. some portion of the outer 

 surface of the glass having probably been dissolved. a result 

 of the constant motion of the water surrounding the instru- 

 ment when in the water-bath. The uncertainty this occa- 

 sions may however be very considerably reduced by com- 

 puting the weight of the pycnometer for every experiment 

 on the assumption that the loss of weight is proportional 

 tu the number of observations; in that case it will hardly 

 influence the 5th decimal. On the basis of this assump- 

 tion. I subsequently computed the following results. intro- 

 ducing here and there slight approximations. which. however. 

 can affect the final result only by occasioning a small error 

 in the 6th decimal. 



Specific Gravitv at ^3 of III 1.01739. 



17.°5 



17."5 

 17.°5 



17:5 

 0° 



of Vn 1.02669. 

 of I 1.02691. 

 of I 1.02845. 



i Pogg. Ann. 153, p. 440. 



- These values were also adopted by Ekman for determining the 

 expansion of the dilatometer which he used when investigating the 

 expansion of sea-water. 



11 Ann. Chim. Phys. [4]. 17. p. 372. 



