that in the presence of neutral salts there is a “primary neutral salt effect” in a negative direction and leave a further explanation until more is known about the general question of primary salt effects. On the other hand, it has been claimed that this anomaly can be removed by assuming that catalysis is due to undissociated acid molecules. This view can be shown to be erroneous, for let Ka be the acid dissociation constant, then

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

K_A=_{γ H γ Cl} C _H C _{Cl/C HCl}.

since the activity coefficient of the undissociated molecule may be taken as unity. The rate of reaction is now given by:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

—d[A]/dt = k[A]C _HCl = k″[A]_γ H γ Cl_{C H} C _Cl

This view, therefore, leads to an equation identical with that which is deduced on the assumption of complete dissociation; in fact, the two postulates cannot be distinguished by experiments at a single temperature.

The temperature coefficient of the reaction may, however, distinguish between the two postulates. Before proceeding further with the argument, it will be necessary to make some assumption as to the number of ternary collisions which can occur in a given time. The number of binary collisions per c.c. per sec. is:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Z₁₂ = n₁ n₂ σ² { 8 π R T (1/M₁ + 1/M₂)}

where M₁ and M₂ are molecular weights and n₁ and n₂ the number of molecules of each species per c.c.

A ternary collision will occur if at the moment of impact the third molecule is within a specific distance r of the complex formed by the first two molecules. The probability of this is ¾ π r³ n₃, so that:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Z₁₂₃ = 4/3 π n₁ n₂ n₃ r³ σ ² { 8 π RT (1/M₁ + 1/M₂)}

where n₃ is the number of molecules of the third kind per c.c.

The number of molecules decomposing per c.c. per sec. is then:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

—dn/dt = 4/3 π n₁ n₂ n₃ r³ σ² {8 π R T (1/M₁ + 1/M₂)} P e ^—E/RT

P being an orientation factor which is probably low for trimolecular reactions. Consequently:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

k₃ = 80π (N_o/1000)² r³ σ² {8 π R T (1/M¹ + 1/M₂)} Pe —^E/RT

The data of Stern (1904) for the polymerisation of benzaldehyde in the presence of cyanide ions are given by:

—13000/RT

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

k₃ = 3.16 × 10⁷ e

Similarly the data of Bailey (1930) on the reaction between ammonia and ethyl malonate are given by:

—8620/RT

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

k₃ = 2-46 × 10⁸ e These reactions can be brought into line with the above equation, assuming reasonable molecular diameters, r = σ = 5 Å, if P is of the order of 10^-5.

These calculations have no claim to accuracy, but they serve to demonstrate that termolecular reactions in solutions can be accounted for, at least qualitatively, the point to be emphasised being that the observed rate is lower than the calculated, as would be expected in view of the peculiar orientation of molecules necessary for such reaction. The decomposition of N-chloracetanilide can be represented by the equation:

—20800/RT

k₃ = 10¹⁴ e (7) whereas the calculated value of the collision factor is of the order of 10¹². Since errors in our theory will tend to give too high a value to this collision factor, we may conclude that our results are not consistent with the hypothesis of ternary collision.

It may be, however, that reaction occurs between anilide molecules and undissociated acid molecules according to the approximate equation:

—d[A]/dt =k [A] C _HCl, in which case k should be capable of calculation by means of equation (6). An approximate estimate of the dissociation constant of hydrochloric acid has already been made (Robinson, 1936) and thence we may calculate the following values for C _HCl in 0.3 M solution at different temperatures whilst the experimental data for the rate of reaction gives values of k as follows:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

These velocity constants can be represented by the equation:

—20800/RT

k = 8 × 10⁶ e (7) with an experimental value for the collision factor of 8 × 10⁶, whilst the value for the collision factor calculated by equation (6) is of the order 10¹³. This implies that only one in about 10⁶ of the activated collisions is effective, but it is encouraging to find that the value from the experimental data is less than the calculated, as would be expected for a reaction of this type; indeed, Rolfe and Hinshel-wood (1934) found a probability factor of 10⁷ for the esterification of methyl alcohol catalysed by undissociated acetic acid molecules. We have therefore a reasonable hypothesis to account for the temperature coefficient of this reaction. We may be tempted to identify these undissociated hydrochloric acid molecules with the “ion pairs” of Bjerrum's association hypothesis, but that such is unlikely may be shown by a consideration of the energy magnitudes involved. The energy liberated when an ion pair is formed is of the order of 2 k T,