My experience supports Dr Sánchez and Dr Kontogiorgos. The relative viscosities (relative efflux times) measured using a Ubbelohde viscometer need to be extrapolated to zero concentration. This can be done with a Ubbelohde viscometer though a capillary suitable for solvent and solution needs to be chosen to give significant different between efflux times for solution and pure solvent. Since the solutions are dilute, precise measurement of times is required and in qualitative appearance the viscosities do not seem different. Even coquette and coaxial cylinder viscometers may lack the precision for the capillary technique. Another factor is that the measurement should involve a slow time because polymer random coils should not be disturbed by flow. Much has been written on sources of error in dilute solution viscometry (see ISO 3104 and ISO 3105). I have mostly used a Ubbelohde viscometer and stopwatch, however currently I use an Schott automatic Ubbelohde viscometer system (Dr Leo Ubbelohde would love this instrument) that has optical measurement of the solution passing graduation marks and automatic dilution between measurements, however these instruments are expensive and useful only for this single task.

Thank you very much for sharing your wisdom on this subject! It has been truly useful.

When the viscosity of the polymer is high, it is necessary to use a rheometer, as capillary viscometers are unable to give valid measures

Would it matter particularly what shear rate the viscosity determinations were performed at? Can anyone suggest a good shear rate for this measurement?

I do not work some time ago, but I remember that the important thing is to work in higher ranges of the scale of torque (from 30%); rheometers manuals are wont explain. It is desirable to use various shear rates (gamma ·) and the value of viscosity is obtained by adjusting the shear stress (tau), depending on gamma ·)

You can use a shear viscometer to find the itrinsic viscosity of dilute polymer solutions and use the Einstein equation to cross check the experimental results. 

Rakesh - I don't understand your point. Do you just mean that the relative viscosity should be a linear function of the polymer concentration?

Theories of polymer solution use the intrinsic viscosity extrapolated to zero shear rate. For not-too-big coils, Ubbelohde is fine, but for highly shear-thinning molecules, xanthan for instance, standard Ubbelohde viscometers will not give correct results. See Guy Chauveteau's amazing classic "Rodlike Polymer Solution Flow through Fine Pores: Influence of Pore Size on Rheological Behavior", J. Rheol. 1982.

However, the data treatment for calculation of the intrinsic viscosity demands very high precision & data from a rotational viscometer, even taking all precautions is unlikely to be good enough. I'd love to see a worked example if someone has succeeded.

To get good data for low viscosities from a stress-controlled rheometer, use a creep protocol. The results are far superior to those obtained with shear rate control as the apparatus does not have to use its feedback loop. However you have to wait for the effect of instrumental inertia to disappear. Expect an asymptotic increase in shear rate over several minutes. I have trick to avoid this, but this note is already long enough. More details on request.

Formally it is possible if data are extrapolated to zero-shear conditions. However, it is much better to work with current Ubbelhode viscometers.

If you use the rheometer, make sure you use a frequency sweep analysis and then you can extrapolated from that your values.. depends on which kind of polymer you use, the increase of conc will increase the shear thinning or shear thickening behavior. this can affect the results obtained from Ubbelhode viscometers. 

Ps: I will not expect a linear relationship.. 

And in fact, the accuracy is rather poor, because extrapolation to zero shear can be cumbersome in many polymer solutions. 

If we have a super sensitive transducer of the rheometer, that would accurately differentiate between the solvent and say a ~ 0.1%  polymer solution, then it is possible, But usually this is not the case.

   Rheometers measure an absolute viscosity, rather than a relative viscosity. To make a determination of intrinsic viscosity, one must measure the viscosity of a solvent and the viscosity of a very dilute solution of polymer and then take the natural log of the ratio of the viscosities.

Intrinsic viscosity determinations require precisions of measurement of relative viscosity of about 1 part per thousand or better. Rheometers are usually much more imprecise measurement devices. They are more in the 1 part per hundred (or less) range. That makes rheometers useless for this purpose.

We  used to build relative viscometers with measurement precisions of relative viscosity directly, and to precisions of 1 part in 30,000. This (at the time 40 years ago) was the best the world had to offer. It let us determine the intrinsic viscosity  to better than 0.3%.

The above link, long winded as it is,  does not point out the importance of experimental details required to attain the precision (and accuracy) of the relative viscosity determination. The most important aspects of the determination all relate to the definition of the determination: ratio of the viscosities of a solvent to a dilute solution of polymer. The practical things that effect this determination of ratios are mostly caused by time fluctuations of a key unspoken variable, Temperature. The change in viscosity with temperature is very large, compared to precision of viscosity determination required for good inherent/intrinsic viscosity determinations. For that reason, a classical setup is a large thermal mass (typically a well stirred tank) with temperature control to approx. .01 centigrade degrees. This minimizes the effect of temperature (approx. .5%/centigrade degree)  on the determination which needs 1 part per thousand for decent data.

Wallace Yau and I invented a flow through relative viscometer which  surpassed the performance of the glass capillary tube setups by more than a factor of 20, and the sample throughput was about 1 sample per 2 minutes. The basic description and background information is found in us patents 4,627,271 and 4,578,990 (Abbott, et al).

It is an easy matter to adapt this principle to chromatographs (e.g. gpc, hplc) and also use these as stand alone viscometers. This invention can also be used as the basis for  excellent rheometers, since it is easy to vary the flow and capillary diameter to vary shear rate over wide ranges, and with outstanding precision.

Common rheometers are usually not referenced measurements, so they are prone to many other experimental variables and have lower precision.

Hi, you can use the Huggins-Kramer extrapolation so all the commetns before I perfectly right. As indicated above, Ubelhonde viscometer is far better for very dilute solutions but sometimes small uncertainties can make the fit a bit sensitive.I found that, when I did this analysis for my set of polymer, using at least one or couple of highly concentrated solutions helped to increase the quality of the fit (to give more weight on the second order term).  Somehow, it corrected for the small inaccuracies you can get when you do very sensitive measures.

Another very good tool is soloman's approximation, if you have only some data from somewhat concentrated solutions. We  usually didn't bother. since the imprecision in relative viscosities using the forced flow dilute solution series capillary viscometer were on the order of 2 parts per 65,000, so we could use rather dilute samples (even of modest molecular weight). 

Check out the following links regarding intrinsic viscosity. You most likely dont need a rheometer to test it. In fact, some issues like solvent evaporation might cause misleading results. A closed system without air-liquid interfaces might give you a more accurate and robust characterization.

http://www.rheosense.com/application/intrinsic-viscosity-with-m-vroc

http://www.rheosense.com/applications/viscosity/intrinsic-viscosity-with-microvisc