Methodology:

Nine records from eight Antarctic expeditions have been utilised in this study. Specifically, these records comprise ship logbooks, meteorological registers, charts and journals. These records were selected based on their fitting of 5 criteria. These criteria are as follows:

1) Records must come from expeditions that took place, at least in part, through the Weddell Sea sector;

2) Records must contain frequent and detailed sea ice observations;

3) Records must contain regular position and time of day. Importantly, the meridian with respect to which ship positions were measured must be clearly stated;

4) Records must be legible in whichever format they are available;

5) Records must be accessible for the general public. In the instance of a record being inaccessible, the relevant authority must have granted permission for the use of this record, for the purpose of this study.

Prior to completing any data extraction, an element of preparation was involved, to efficiently analyse and extract relevant data. Firstly, the journal kept by Dumont D'Urville aboard the 'Astrolabe' required translating from French to English before any information could be extracted. A translated account of this journal is available (Rosenman, 1987) however, to include information that may have been omitted in this translated edition, the original journal was utilised. Then, in analysing historical records, it is important to understand exactly what kind of information is held within each record. Therefore, familiarisation with the format of ship logbooks and meteorological registers, as well as the nautical language typically found in these records, was an important first stage.

Additionally, prior to extracting information regarding sea ice observations the various terminology used in each record was analysed and established as being representative of sea ice or not. Within each record a variety of terms are used in place of sea ice, which was only identified once in one journal account. Interpretation of these various terms can be challenging due to the often-ambiguous manner in which they are used. Through context it was possible to deduce the meaning of different terms and thus distinguish between observations relating to sea ice and other ice forms, such as icebergs or shelf ice. In cases where the meaning can be clearly deduced, the term has been allocated its relevant meaning ('sea ice' or 'not sea ice'). It should be noted that where the use of a term was ambiguous, it was classified as 'not sea ice', to minimize error.

From each record, where a sea ice observation was logged five key points were noted.

These points are as follows:

1) The time of day (usually Noon) and ship's position at the time of the observation;

2) The date on which the observation was made;

3) A detailed description of the sea ice observed;

4) The term/s used to represent sea ice in the observation;

5) Any remarks made by the author comparing their observations of sea ice with observations made on previous expeditions.

Following the data extraction, it was necessary to make certain adjustments, to make the data uniform where necessary and improve the data accuracy. Firstly, longitudes logged on board the Astrolabe were done so with respect to the Paris Meridian, whereas all other longitudes were measured with respect to the Greenwich Meridian. The Paris Meridian is located approximately 2°20'14'' east of the Greenwich Meridian, therefore all longitudes measured from the Astrolabe were adjusted accordingly.

Next, the method of recording ship position was noted upon analysing each record. It is important to know exactly how ship positions were recorded, to estimate the accuracy with which this was done. To validate the accuracy of these positions each route was divided into 'legs'; each leg comprising a section of the route between landmarks (fo...(7)