A relatively constant molecularweight value is obtained at comparatively low viscosities when benzene is usedas the cryoscopic solvent. Under the experimental conditions employed in thethermal polymerization of the oil, this value reaches approximately.200 at aviscosity of 1700 centipoises and does not change appreciably on reaching19,000 centipoises. For the oxidized oil the molecular weight value approaches1400 and again is fairly constant in the viscosity range of 1200 to 3500centipoises.

Molecular weight investigations byCaldwell and Mattiello (2)on polymerized linseed oil using nitrobenzene as thecryoscopic solvent have shown that oils heat-bodied at different temperaturesapproach different constant molecular weights at relatively low viscosities. Thus,at a polymerization temperature of 329.4℃ .the value was approximately 1550; at 304.4℃ ，1300;and at 287.7℃ 1000.Our experiments indicate that benzeneyields similar results. These constant molecular weight values, which extendover a wide viscosity range, indicate that any theoretical significance ofthese determinations in benzene would be questionable.

The use of cyclohexane as solventpresents an entirely different problem. There tile molecular weight increasesrapidly as the viscosity increases, and the polymerized oil at approximately2500 cps. approaches an average molecular weight of the trimer stage. For theoxidized oil the tetramer stage is approached at a viscosity of 2400 cps. Thevalue continues to increase slowly at the higher viscosities and apparentlydoes not become constant in the liquid phase. Above approximately 2500centipoises the method cannot be employed for oxidized oils because ofinsolubility. The question of whether results obtained with cyclohexane aretrue mean molecular weight readings is still open for discussion. It is believedthat if values could be duplicated by using different cryoscopic solvents morepositive statements concerning the true nature of the mean molecular weightcould be made.

Nevertheless, theoretical andactual considerations lead to the assumption that cyclohexane is an almost idealsolvent for cryoscopic measurements. Four facts may be pointed out in supportof this statement:

1.Molecular weights of raw oilagree closely with theoretical calculations.

2.Different concentrations of oilin cyclohexane give nearly identical molecular weight readings. In benzene itwas found that in only some instances arc good check determinations obtained atdifferent concentrations. Mostly, however, the molecular weight values inbenzene decrease as higher concentrations of solute are used.

3.Cyclohexane,as opposed to mostother solvents tried, is a chemically and physically inert substance. Associationwith similar substances or with the solute is minimized since the compound doesnot possess valency stresses. The material is an excellent solute forpolymerized oil and for oxidized oil up to about 2500 centipoises, and it has asharp freezing point in the pure state.

4.The molecular weights obtainedfor polymerized oils at high viscosities or near the gel point agree closelywith the theoretical number average value predicted statistically by Flory'stheory for three dimensional polymers(3).According to this theory, a gel shouldbe obtainable with an oil polymer having an average molecular weightapproximately three times that of the original oil. In the case of the polymerizedoil near the gel point this factor is about 3.5.

It is illustrated in Figure 1that for any given viscosity the oxidized oil has a higher molecular weightthan the polymerized oil. That is true in cyclohexane as well as in benzene. Theseresults however must be regarded with some caution. They are representative ofthe oils processed under the conditions outlined previously and consequently nogeneralization should be attempted. It is conceivable that differentexperimental conditions could affect the reactions taking place during theoxidation or polymerization and therefore could alter the molecularweight-viscosity relationship.

Dielectric
Constant

The method for the determinationof the dielectric, constant and dielectric theory have been fully described anddiscussed by Hazlehurst(6).

The experimental values(Figure2)show that the dielectric constant rises from 3.22 to 5.14 as oxidationproceeds to a viscosity of about 3500 cps. This has enabled us to exercisecontinuous control over the oxidation of oils since a definite relationship existsbetween the dielectric constant and the amount of oxygen in the oil for anygiven set of conditions. Moreover, the approximate amount of oxygen in any unknownoil sample can readily and simply be estimated f r o m its dielectric constantwithout resorting to laborious and time consuming combustion analyses. It is tobe remembered however that for different oxidation conditions, especiallytemperature, the dielectric-oxygen content curve will v a r y somewhat. Therelationship of oxygen content to dielectric constant for this batch is shownin Figure 3.

Cryoscopic determinations witheyelohexane have been further investigated and extended in our laboratory forthermally polymerized and mechanically oxidized raw linseed oil and forpurposes of comparison determinations have also been made in benzene.