### Animal preparation

This study was approved by the Animal Experimentation Ethics Committee of the Howard Florey Institute, Melbourne, Australia. Experiments were performed on eight conscious adult Merino ewes weighing between 25 to 40 kg, housed in individual metabolic cages.

On the day prior to the experiment, an arterial Tygon catheter and internal jugular venous polythene catheter were inserted for blood sampling and fluid infusion, respectively.

### Experimental protocol

Animals were randomly assigned to either sodium chloride (150 mmol Na and 150 mmol Cl) or sodium octanoate (150 mmol Na, 100 mmol chloride, and 50 mmol octanoate). The solutions were provided in indistinguishable glass bottles prepared for the experiment by CSL Bioplasma (CSL, Melbourne, Australia). Thus, assignment was random and concealed, and the investigator infusing the fluid was blinded to the type of fluid. The following day, the animal was assigned to the other fluid in a crossover design, such that those animals first assigned to sodium chloride then received sodium octanoate the next day and those first assigned to sodium octanoate then received sodium chloride the next day.

Baseline blood samples were taken for blood gas analysis (ABL800 Blood Gas Analyzer, Radiometer, Copenhagen, Denmark) and measurement of serum magnesium, phosphate, and albumin levels (SYNCHRON LX® System Beckmann Coulter Inc., Fullerton, CA, USA). After such baseline measurements, animals received a rapid intravenous infusion (over 30 min) of 1 L of trial solution. Additional blood samples were then collected at 0.5, 1, 2, 4, and 6 h after the start of the infusion.

### Calculations for the interpretation of quantitative acid–base analysis

Quantitative biophysical analysis of the results was performed with the Stewart approach [10] as modified by Figge et al. [11] to take into account the effects of other factors like albumin level. This method first calculates the apparent strong ion difference (SIDa):

\begin{array}{c}\hfill \mathrm{SIDa}=\left[{\mathrm{Na}}^{+}\right]+\left[{\mathrm{K}}^{+}\right]+\left[{\mathrm{Mg}}^{2+}\right]+\left[{\mathrm{Ca}}^{2+}\right]\u2012\left[{\mathrm{Cl}}^{\u2012}\right]\u2012\left[\mathrm{Lactate}\right]\hfill \\ \hfill \left(\mathrm{all}\phantom{\rule{0.24em}{0ex}}\mathrm{concentrations}\phantom{\rule{0.24em}{0ex}}\mathrm{in}\phantom{\rule{0.24em}{0ex}}\mathrm{mEq}/\mathrm{L}\right)\hfill \end{array}

However, the role of weak acids (carbon dioxide, albumin, and phosphate) in the balance of electrical charges in plasma water is not taken into account in this equation. The effective strong ion difference (SIDe) was thus calculated according to Figge et al. [11] as follows:

\begin{array}{c}\hfill \mathrm{SIDe}=2{.46\times 10}^{\left(\mathrm{pH}-8\right)}\times p{\mathrm{CO}}_{2}+\left[\mathrm{Albumin}\right]\times \left(0.12\times \mathrm{pH}-0.631\right)+\left[\mathrm{Phosphate}\right]\times \left(0.309\right)\times \mathrm{pH}-0.469.\hfill \\ \hfill \left(p{\mathrm{CO}}_{2}\phantom{\rule{0.24em}{0ex}}\mathrm{in}\phantom{\rule{0.24em}{0ex}}\mathrm{mmHg},\mathrm{albumin}\phantom{\rule{0.24em}{0ex}}\mathrm{in}\phantom{\rule{0.24em}{0ex}}\mathrm{g}/\mathrm{L},\mathrm{and}\phantom{\rule{0.24em}{0ex}}\mathrm{phosphate}\phantom{\rule{0.24em}{0ex}}\mathrm{in}\phantom{\rule{0.24em}{0ex}}\mathrm{mmol}/\mathrm{l}\right)\hfill \end{array}

Without unmeasured charges, SIDa - SIDe should be equal to zero (electrical charge neutrality) once weak acids are quantitatively taken into account. If a gap exists between SIDa and SIDe, then unmeasured anions (e.g., sulfate, keto acids, citrate, pyruvate, acetate, and gluconate) must be present to explain this gap which is termed the strong ion gap (SIG): SIG = SIDa - SIDe.

### Statistical analysis

To ascertain if NS was significantly different from octanoate solution (OS) over the 6-h period or if NS behaved differently over time, we used repeated measures analysis of variance model fitting main effects for group (NS or OS), time (two times using 0 and 6 h), and an interaction between group and time to determine if the two groups behave differently over time. Modeling was performed using the PROC Mixed procedure in SAS version 9.2 (SAS Institute Inc., Cary, NC, USA) with the six individual sheep treated as random effects. Nine outcome variables were considered (pH, base excess, hematocrit, sodium, chloride, bicarbonate, SIDa, SIDe, SIG), meaning nine models were constructed. Results have been reported as least square means (95% confidence intervals).