Most vegetable oils can be distinguished from one another by their colour and taste: corn oil and sunflower oil, which you mentioned, are physically distinguishable from one another in this way. If you suspect your oil is adulterated with other oils, you don't need a chemical test.

All you need is a very accurate thermometer, and apparatus to test for the smoke point of the oil. http://www.culinary-yours.com/frying_oil.html

I suggest you

• Buy a "Pounds per Gallon" cup to see density and specific gravity.
• Also have pure oils in test tubes for color comparison as the other blogger stated.
• If you can afford GCMS testing will be ideal

That mixture is not a health hazard, but it is frustrating to pay for something that is adulterated.

If you are so concerned about purity, then you're probably a chef and can afford buying only reputaded brands, which will state the prduct contents on the label, you can trust them, they will not risk their hard gained prestige and certainly don't want to get sued.

Yahlasit

I have seen a recent paper in energy and fuel periodical published by the ACS titled Identification of the Biodiesel Source Using an Electronic Nose. I don't know the price of such an electronic nose but in the text they refer to «Smiths Detection Cyranose 320 e-nose» and they claimed that this technique is cheaper than traditionnal method. The drawback of using an e-nose is that you have to train it in order to recognise the sample and that the reading must be done in the same conditions that the training one. Also there is no profond chemical interpretation possible. However, depending on the task you need, I think it can be an option for you because in this particular paper they claimed that the electronic nose was also tested to recognize the origin of the biodiesel in B20 blends, meaning that it can recognize different ratio of different FAME (faty methyl ester).

Portable Raman could do this. It's spectrophotometry with a nifty nose. Unfortunately, expensive to acquire - maybe $10,000 for a bare bones model if you're lucky - but a great tool to have if you need to test on a regular basis. (I want one! for everything)