1. A new simple method is described for quantitating the adhesiveness of circulating polymorphonuclear leucocytes, or granulocytes, to the walls of blood vessels. The cheek pouch of anaesthetized hamsters or a small part of the mesentery of anaesthetized mice were prepared for continuous microscopic observation of selected venules. Those granulocytes which moved sufficiently slowly to be individually visible were counted for 1 or 2 min periods as they rolled past a selected point on one side of a vessel. The velocity distribution of these cells was determined by analysing films. Films were used also to measure mean blood flow velocity in the venules by observing embolizing platelet thrombi induced by the iontophoretic application of adenosine diphosphate. Emigration of granulocytes into the tissues was quantitated by enumerating them in standard areas of stained histological sections.

2. In control experiments with hamster cheek pouch venules, the rolling granulocyte count usually passed through a maximum shortly after the preparation was set up and then fell to a low constant value. In mouse mesentery venules the count remained at a low approximately constant value from the beginning for at least 3 hr.

3. The mean velocity of blood flow in the venules was between 900 and 200 μ/sec. All rolling granulocytes moved much more slowly; in hamster cheek pouch venules the mean velocity was about 20 μ/sec and in mouse mesentery venules about 10 μ/sec. Around these means the velocity distribution of individual cells was narrow.

4. Rolling of granulocytes was abolished by superfusing ethylenediamine tetra‐acetate (EDTA, 0·1 M) suggesting that the phenomenon depends on calcium or magnesium ions.

5. Agents were applied locally to the observed venules. Human serum albumin, trypsin or histamine in high concentrations did not affect the rolling granulocyte count.

6. The rolling granulocyte count was increased during the application of Hammarsten casein or Escherichia coli culture filtrate which are chemotactic to granulocytes in vitro. These agents did not cause alterations in mean blood flow velocity in the observed venules which might have accounted for the effect on the rolling granulocyte counts. When E. coli culture filtrate was perfused through mouse intestine the increase in rolling granulocyte count in the draining venous blood was proportional to the logarithm of the concentration of filtrate.

7. The rolling granulocyte count was also increased by the local application of plasma globulin permeability factor or lymph node permeability factor.

8. Granulocyte counts in standard histological sections showed no significant increases in control preparations but considerable increases following the application of Hammarsten casein.