Source data must be in columns as: [CaseName] [V1] [V2] ... (each object in a separate row, column separator can be space, tab or semicolon, decimal separator can be either point or comma). The case name can be either object name or a numeric ID. First row can contain variable names. All variables must be numeric. k-means uses possibly distant cases as cluster seeds.
See also One dimensional clustering.
Article in Wikipedia. A presentation by M. Pardo. Exercises: 3665, 3666, 3668, 4041 and 4042.